Merge commit '52e5f0a5c1' into concedo_experimental

# Conflicts:
#	examples/gen-docs/gen-docs.cpp
#	examples/lookup/lookup-create.cpp
#	examples/lookup/lookup-stats.cpp
#	examples/lookup/lookup.cpp
#	examples/speculative-simple/speculative-simple.cpp
#	examples/speculative/speculative.cpp
#	ggml/src/CMakeLists.txt
#	ggml/src/ggml-cann/aclnn_ops.cpp
#	ggml/src/ggml-cann/aclnn_ops.h
#	ggml/src/ggml-cann/ggml-cann.cpp
#	ggml/src/ggml-rpc/ggml-rpc.cpp
#	ggml/src/ggml-vulkan/ggml-vulkan.cpp
#	ggml/src/ggml-webgpu/ggml-webgpu-shader-lib.hpp
#	ggml/src/ggml-webgpu/ggml-webgpu.cpp
#	ggml/src/ggml-webgpu/wgsl-shaders/binary.wgsl
#	ggml/src/ggml-webgpu/wgsl-shaders/get_rows.wgsl
#	ggml/src/ggml-webgpu/wgsl-shaders/mul_mat_decls.tmpl
#	ggml/src/ggml-webgpu/wgsl-shaders/mul_mat_vec.wgsl
#	ggml/src/ggml-webgpu/wgsl-shaders/rms_norm_mul.wgsl
#	ggml/src/ggml-webgpu/wgsl-shaders/ssm_scan.wgsl
#	tests/test-arg-parser.cpp
#	tests/test-backend-ops.cpp
#	tests/test-chat.cpp
#	tests/test-reasoning-budget.cpp
#	tools/llama-bench/llama-bench.cpp
#	tools/rpc/rpc-server.cpp
#	tools/server/webui/src/lib/components/app/chat/ChatScreen/ChatScreen.svelte
#	tools/server/webui/src/lib/components/app/chat/ChatSidebar/ChatSidebar.svelte
#	tools/server/webui/src/routes/(chat)/+page.svelte

common/arg.h

@@ -25,7 +25,8 @@ struct common_arg {
     const char * value_hint_2 = nullptr; // for second arg value
     const char * env          = nullptr;
     std::string help;
-    bool is_sparam = false; // is current arg a sampling param?
+    bool is_sampling = false; // is current arg a sampling param?
+    bool is_spec = false; // is current arg a speculative decoding param?
     bool is_preset_only = false; // is current arg preset-only (not treated as CLI arg)
     void (*handler_void)   (common_params & params) = nullptr;
     void (*handler_string) (common_params & params, const std::string &) = nullptr;
@@ -74,7 +75,8 @@ struct common_arg {
     common_arg & set_examples(std::initializer_list<enum llama_example> examples);
     common_arg & set_excludes(std::initializer_list<enum llama_example> excludes);
     common_arg & set_env(const char * env);
-    common_arg & set_sparam();
+    common_arg & set_sampling();
+    common_arg & set_spec();
     common_arg & set_preset_only();
     bool in_example(enum llama_example ex);
     bool is_exclude(enum llama_example ex);

common/common.cpp

@@ -76,7 +76,7 @@ common_time_meas::~common_time_meas() {
 // CPU utils
 //
 
-int32_t cpu_get_num_physical_cores() {
+int32_t common_cpu_get_num_physical_cores() {
 #ifdef __linux__
     // enumerate the set of thread siblings, num entries is num cores
     std::unordered_set<std::string> siblings;
@@ -191,11 +191,11 @@ static int cpu_count_math_cpus(int n_cpu) {
 /**
  * Returns number of CPUs on system that are useful for math.
  */
-int32_t cpu_get_num_math() {
+int32_t common_cpu_get_num_math() {
 #if defined(__x86_64__) && defined(__linux__) && !defined(__ANDROID__)
     int n_cpu = sysconf(_SC_NPROCESSORS_ONLN);
     if (n_cpu < 1) {
-        return cpu_get_num_physical_cores();
+        return common_cpu_get_num_physical_cores();
     }
     if (is_hybrid_cpu()) {
         cpu_set_t affinity;
@@ -208,7 +208,7 @@ int32_t cpu_get_num_math() {
         }
     }
 #endif
-    return cpu_get_num_physical_cores();
+    return common_cpu_get_num_physical_cores();
 }
 
 // Helper for setting process priority
@@ -269,7 +269,7 @@ bool set_process_priority(enum ggml_sched_priority prio) {
 //
 
 
-void postprocess_cpu_params(cpu_params& cpuparams, const cpu_params* role_model) {
+void postprocess_cpu_params(common_cpu_params & cpuparams, const common_cpu_params * role_model) {
     int32_t n_set = 0;
 
     if (cpuparams.n_threads < 0) {
@@ -277,7 +277,7 @@ void postprocess_cpu_params(cpu_params& cpuparams, const cpu_params* role_model)
         if (role_model != nullptr) {
             cpuparams = *role_model;
         } else {
-            cpuparams.n_threads = cpu_get_num_math();
+            cpuparams.n_threads = common_cpu_get_num_math();
         }
     }
 
@@ -1527,7 +1527,7 @@ struct llama_context_params common_context_params_to_llama(const common_params &
     return cparams;
 }
 
-struct ggml_threadpool_params ggml_threadpool_params_from_cpu_params(const cpu_params & params) {
+struct ggml_threadpool_params ggml_threadpool_params_from_cpu_params(const common_cpu_params & params) {
     struct ggml_threadpool_params tpp;
 
     ggml_threadpool_params_init(&tpp, params.n_threads); // setup the defaults

common/common.h

@@ -55,7 +55,7 @@ struct common_control_vector_load_info;
 // CPU utils
 //
 
-struct cpu_params {
+struct common_cpu_params {
     int      n_threads                   = -1;
     bool     cpumask[GGML_MAX_N_THREADS] = {false}; // CPU affinity mask.
     bool     mask_valid                  = false;   // Default: any CPU
@@ -64,8 +64,8 @@ struct cpu_params {
     uint32_t poll                        = 50;      // Polling (busywait) level (0 - no polling, 100 - mostly polling)
 };
 
-int32_t cpu_get_num_physical_cores();
-int32_t cpu_get_num_math();
+int32_t common_cpu_get_num_physical_cores();
+int32_t common_cpu_get_num_math();
 
 //
 // Common params
@@ -298,34 +298,19 @@ struct common_params_model {
 
 struct common_ngram_mod;
 
-struct common_params_speculative {
-    common_speculative_type type = COMMON_SPECULATIVE_TYPE_NONE; // type of speculative decoding
+// draft-model-based speculative decoding parameters
+struct common_params_speculative_draft {
+    int32_t n_max = 16; // maximum number of tokens to draft during speculative decoding
+    int32_t n_min = 0;  // minimum number of draft tokens to use for speculative decoding
 
-    // general-purpose speculative decoding parameters
+    float p_split = 0.1f;  // speculative decoding split probability
+    float p_min   = 0.75f; // minimum speculative decoding probability (greedy)
 
-    int32_t n_max   = 16; // maximum number of tokens to draft during speculative decoding
-    int32_t n_min   = 0;  // minimum number of draft tokens to use for speculative decoding
-    float   p_split = 0.1f; // speculative decoding split probability
-    float   p_min   = 0.75f; // minimum speculative decoding probability (greedy)
+    common_params_model mparams;
 
-    // ngram-based speculative decoding
+    llama_model * model = nullptr; // a llama_model that can be shared by multiple speculative contexts
 
-    uint16_t ngram_size_n   = 12; // ngram size for lookup
-    uint16_t ngram_size_m   = 48; // mgram size for speculative tokens
-    uint16_t ngram_min_hits = 1; // minimum hits at ngram/mgram lookup for mgram to be proposed
-
-    std::shared_ptr<common_ngram_mod> ngram_mod;
-
-    std::string lookup_cache_static;  // path of static ngram cache file for lookup decoding           // NOLINT
-    std::string lookup_cache_dynamic; // path of dynamic ngram cache file for lookup decoding          // NOLINT
-
-    // draft-model speculative decoding
-
-    struct common_params_model mparams_dft;
-
-    llama_model * model_dft = nullptr; // a llama_model that can be shared by multiple speculative contexts
-
-    llama_context_params cparams_dft; // these are the parameters for the draft llama_context
+    llama_context_params cparams; // these are the parameters for the draft llama_context
 
     int32_t n_ctx        = 0;  // draft context size
     int32_t n_gpu_layers = -1; // number of layers to store in VRAM for the draft model (-1 - use default)
@@ -333,25 +318,60 @@ struct common_params_speculative {
     ggml_type cache_type_k = GGML_TYPE_F16; // KV cache data type for the K
     ggml_type cache_type_v = GGML_TYPE_F16; // KV cache data type for the V
 
-    struct cpu_params cpuparams;
-    struct cpu_params cpuparams_batch;
+    common_cpu_params cpuparams;
+    common_cpu_params cpuparams_batch;
 
     std::vector<ggml_backend_dev_t> devices; // devices to use for offloading
 
     std::vector<std::pair<std::string, std::string>> replacements; // main to speculative model replacements
     std::vector<llama_model_tensor_buft_override> tensor_buft_overrides;
+};
+
+struct common_params_speculative_ngram_mod {
+    int32_t n_match = 24;
+
+    int32_t n_max = 64;
+    int32_t n_min = 48;
+
+    // shared instance of the ngram container for all speculative decoding contexts
+    std::shared_ptr<common_ngram_mod> obj;
+};
+
+struct common_params_speculative_ngram_map {
+    uint16_t size_n   = 12; // ngram size for lookup
+    uint16_t size_m   = 48; // mgram size for speculative tokens
+    uint16_t min_hits = 1;  // minimum hits at ngram/mgram lookup for mgram to be proposed
+};
+
+struct common_params_speculative_ngram_cache {
+    std::string lookup_cache_static;  // path of static ngram cache file for lookup decoding
+    std::string lookup_cache_dynamic; // path of dynamic ngram cache file for lookup decoding
+};
+
+struct common_params_speculative {
+    // TODO: become a vector in order to support "chains of speculators"
+    common_speculative_type type = COMMON_SPECULATIVE_TYPE_NONE;
+
+    common_params_speculative_draft draft;
+
+    common_params_speculative_ngram_mod ngram_mod;
+    common_params_speculative_ngram_map ngram_simple;
+    common_params_speculative_ngram_map ngram_map_k;
+    common_params_speculative_ngram_map ngram_map_k4v;
+
+    common_params_speculative_ngram_cache ngram_cache;
 
     bool has_dft() const {
-        return !mparams_dft.path.empty() || !mparams_dft.hf_repo.empty();
+        return !draft.mparams.path.empty() || !draft.mparams.hf_repo.empty();
     }
 };
 
 struct common_params_vocoder {
     struct common_params_model model;
 
-    std::string speaker_file = ""; // speaker file path                                      // NOLINT
+    std::string speaker_file; // speaker file path
 
-    bool use_guide_tokens = false; // enable guide tokens to improve TTS accuracy            // NOLINT
+    bool use_guide_tokens = false; // enable guide tokens to improve TTS accuracy
 };
 
 struct common_params_diffusion {
@@ -434,8 +454,8 @@ struct common_params {
 
     enum llama_split_mode split_mode = LLAMA_SPLIT_MODE_LAYER; // how to split the model across GPUs
 
-    struct cpu_params cpuparams;
-    struct cpu_params cpuparams_batch;
+    common_cpu_params cpuparams;
+    common_cpu_params cpuparams_batch;
 
     ggml_backend_sched_eval_callback cb_eval = nullptr;
     void * cb_eval_user_data                 = nullptr;
@@ -679,7 +699,7 @@ std::string common_params_get_system_info(const common_params & params);
 
 bool parse_cpu_range(const std::string & range, bool(&boolmask)[GGML_MAX_N_THREADS]);
 bool parse_cpu_mask(const std::string & mask, bool(&boolmask)[GGML_MAX_N_THREADS]);
-void postprocess_cpu_params(cpu_params & cpuparams, const cpu_params * role_model = nullptr);
+void postprocess_cpu_params(common_cpu_params & cpuparams, const common_cpu_params * role_model = nullptr);
 bool set_process_priority(enum ggml_sched_priority prio);
 
 //
@@ -847,7 +867,7 @@ common_init_result_ptr common_init_from_params(common_params & params);
 
 struct llama_model_params     common_model_params_to_llama  (      common_params & params);
 struct llama_context_params   common_context_params_to_llama(const common_params & params);
-struct ggml_threadpool_params ggml_threadpool_params_from_cpu_params(const cpu_params & params);
+struct ggml_threadpool_params ggml_threadpool_params_from_cpu_params(const common_cpu_params & params);
 
 // clear LoRA adapters from context, then apply new list of adapters
 void common_set_adapter_lora(struct llama_context * ctx, std::vector<common_adapter_lora_info> & lora);

common/preset.cpp

@@ -43,7 +43,7 @@ static std::set<std::string> get_remote_preset_whitelist(const std::map<std::str
     for (const auto & it : key_to_opt) {
         const std::string & key = it.first;
         const common_arg & opt = it.second;
-        if (allowed_options.find(key) != allowed_options.end() || opt.is_sparam) {
+        if (allowed_options.find(key) != allowed_options.end() || opt.is_sampling) {
             allowed_keys.insert(key);
             // also add variant keys (args without leading dashes and env vars)
             for (const auto & arg : opt.get_args()) {

common/reasoning-budget.cpp

@@ -122,6 +122,20 @@ static void common_reasoning_budget_accept(struct llama_sampler * smpl, llama_to
             }
             break;
         case REASONING_BUDGET_DONE:
+            // Re-arm on a new start tag: some models emit multiple <think> blocks
+            // per response, and each should get a fresh budget window.
+            if (ctx->start_matcher.advance(token)) {
+                ctx->state = REASONING_BUDGET_COUNTING;
+                ctx->remaining = ctx->budget;
+                ctx->end_matcher.reset();
+                LOG_INF("reasoning-budget: re-activated on new start tag, budget=%d tokens\n", ctx->budget);
+
+                if (ctx->remaining <= 0) {
+                    ctx->state = REASONING_BUDGET_FORCING;
+                    ctx->force_pos = 0;
+                    LOG_INF("reasoning-budget: budget=0, forcing immediately\n");
+                }
+            }
             break;
     }
 }

common/speculative.cpp

@@ -151,6 +151,9 @@ struct common_speculative_state {
             llama_tokens & result) = 0;
 
     virtual void accept(uint16_t n_accepted) = 0;
+
+    virtual int32_t n_max(const common_params_speculative & params) const = 0;
+    virtual int32_t n_min(const common_params_speculative & params) const = 0;
 };
 
 struct common_speculative_checkpoint {
@@ -296,6 +299,8 @@ struct common_speculative_state_draft : public common_speculative_state {
             const llama_tokens & prompt_tgt,
             llama_token id_last,
             llama_tokens & result) override {
+        const auto & sparams = params.draft;
+
         auto * spec = this;
 
         auto & batch      = spec->batch;
@@ -309,7 +314,7 @@ struct common_speculative_state_draft : public common_speculative_state {
         int reuse_i = 0; // index of part to be reused in prompt_dft
         int reuse_n = 0; // length of part to be reused in prompt_dft
 
-        const int n_ctx = llama_n_ctx(ctx_dft) - params.n_max;
+        const int n_ctx = llama_n_ctx(ctx_dft) - sparams.n_max;
 
         llama_tokens prompt_cnv;
         if (!spec->vocab_cmpt) {
@@ -367,7 +372,7 @@ struct common_speculative_state_draft : public common_speculative_state {
         }
 
         result.clear();
-        result.reserve(params.n_max);
+        result.reserve(sparams.n_max);
 
         bool needs_ckpt = use_ckpt && prompt_dft.size() > 0;
         if (reuse_n == 0 || (use_ckpt && reuse_i > 0)) {
@@ -380,7 +385,7 @@ struct common_speculative_state_draft : public common_speculative_state {
                 for (int i = reuse_i + reuse_n + 1; i < (int) prompt_dft.size(); ++i) {
                     result.push_back(prompt_dft[i]);
 
-                    if (params.n_max <= (int) result.size()) {
+                    if (sparams.n_max <= (int) result.size()) {
                         break;
                     }
                 }
@@ -473,7 +478,7 @@ struct common_speculative_state_draft : public common_speculative_state {
         common_sampler_reset(smpl);
 
         // sample n_draft tokens from the draft model
-        for (int i = 0; i < params.n_max; ++i) {
+        for (int i = 0; i < sparams.n_max; ++i) {
             common_batch_clear(batch);
 
             common_sampler_sample(smpl, ctx_dft, 0, true);
@@ -492,12 +497,12 @@ struct common_speculative_state_draft : public common_speculative_state {
 
             result.push_back(id);
 
-            if (params.n_max <= (int) result.size()) {
+            if (sparams.n_max <= (int) result.size()) {
                 break;
             }
 
             // only collect very high-confidence draft tokens
-            if (cur_p->data[0].p < params.p_min) {
+            if (cur_p->data[0].p < sparams.p_min) {
                 break;
             }
 
@@ -518,10 +523,14 @@ struct common_speculative_state_draft : public common_speculative_state {
             detokenized = replace_to_tgt(detokenized);
             LOG_DBG("draft->main detokenized string: '%s'\n", detokenized.c_str());
             result = common_tokenize(ctx_tgt, detokenized, false, true);
-            if (result.size() > (size_t)params.n_max) {
-                result.resize(params.n_max);
+            if (result.size() > (size_t) sparams.n_max) {
+                result.resize(sparams.n_max);
             }
         }
+
+        if (result.size() < (size_t) sparams.n_min) {
+            result.clear();
+        }
     }
 
     void accept(uint16_t n_accepted) override {
@@ -529,6 +538,14 @@ struct common_speculative_state_draft : public common_speculative_state {
         GGML_UNUSED(n_accepted);
     }
 
+    int32_t n_max(const common_params_speculative & params) const override {
+        return params.draft.n_max;
+    }
+
+    int32_t n_min(const common_params_speculative & params) const override {
+        return params.draft.n_min;
+    }
+
     std::string replace_to_dft(const std::string & input) const {
         std::string result = input;
 
@@ -581,6 +598,14 @@ struct common_speculative_state_eagle3 : public common_speculative_state {
         // noop
         GGML_UNUSED(n_accepted);
     }
+
+    int32_t n_max(const common_params_speculative & params) const override {
+        return params.draft.n_max;
+    }
+
+    int32_t n_min(const common_params_speculative & params) const override {
+        return params.draft.n_min;
+    }
 };
 
 // state of self-speculation (simple implementation, not ngram-map)
@@ -610,19 +635,27 @@ struct common_speculative_state_ngram_simple : public common_speculative_state {
         // noop
         GGML_UNUSED(n_accepted);
     }
+
+    int32_t n_max(const common_params_speculative & /*params*/) const override {
+        return config.size_mgram;
+    }
+
+    int32_t n_min(const common_params_speculative & /*params*/) const override {
+        return config.size_mgram;
+    }
 };
 
 struct common_speculative_state_ngram_map_k : public common_speculative_state {
     // draft ngram map for speculative decoding without draft model
-    common_ngram_map map;
+    common_ngram_map config;
 
     common_speculative_state_ngram_map_k(
             enum common_speculative_type type,
-            common_ngram_map map)
-        : common_speculative_state(type), map(std::move(map)) {}
+            common_ngram_map config)
+        : common_speculative_state(type), config(std::move(config)) {}
 
     void begin(const llama_tokens & prompt) override {
-        common_ngram_map_begin(map, prompt);
+        common_ngram_map_begin(config, prompt);
     }
 
     void draft(
@@ -630,12 +663,20 @@ struct common_speculative_state_ngram_map_k : public common_speculative_state {
             const llama_tokens & prompt_tgt,
             llama_token id_last,
             llama_tokens & result) override {
-        common_ngram_map_draft(map, prompt_tgt, id_last, result);
+        common_ngram_map_draft(config, prompt_tgt, id_last, result);
         GGML_UNUSED(params);
     }
 
     void accept(uint16_t n_accepted) override {
-        common_ngram_map_accept(map, n_accepted);
+        common_ngram_map_accept(config, n_accepted);
+    }
+
+    int32_t n_max(const common_params_speculative & /*params*/) const override {
+        return config.size_value;
+    }
+
+    int32_t n_min(const common_params_speculative & /*params*/) const override {
+        return config.size_value;
     }
 };
 
@@ -692,7 +733,7 @@ struct common_speculative_state_ngram_mod : public common_speculative_state {
             const llama_tokens & prompt_tgt,
             llama_token id_last,
             llama_tokens & result) override {
-        GGML_UNUSED(params);
+        const auto & sparams = params.ngram_mod;
 
         n_draft_last = 0;
 
@@ -712,16 +753,16 @@ struct common_speculative_state_ngram_mod : public common_speculative_state {
             i_last = cur_len - n;
         }
 
-        result.resize(n + params.n_max);
+        result.resize(n + sparams.n_max);
         for (size_t i = 0; i < n - 1; ++i) {
             result[i] = prompt_tgt[cur_len - n + 1 + i];
         }
         result[n - 1] = id_last;
 
-        for (int i = 0; i < params.n_max; ++i) {
+        for (int i = 0; i < sparams.n_max; ++i) {
             const llama_token token = mod.get(result.data() + i);
             if (token == common_ngram_mod::EMPTY) {
-                if (i < params.n_min) {
+                if (i < sparams.n_min) {
                     result.clear();
                     return;
                 }
@@ -764,6 +805,14 @@ struct common_speculative_state_ngram_mod : public common_speculative_state {
             }
         }
     }
+
+    int32_t n_max(const common_params_speculative & params) const override {
+        return params.ngram_mod.n_max;
+    }
+
+    int32_t n_min(const common_params_speculative & params) const override {
+        return params.ngram_mod.n_min;
+    }
 };
 
 struct common_speculative_state_ngram_cache : public common_speculative_state {
@@ -857,6 +906,14 @@ struct common_speculative_state_ngram_cache : public common_speculative_state {
         // TODO: noop
         GGML_UNUSED(n_accepted);
     }
+
+    int32_t n_max(const common_params_speculative & /*params*/) const override {
+        return n_draft;
+    }
+
+    int32_t n_min(const common_params_speculative & /*params*/) const override {
+        return 0;
+    }
 };
 
 struct common_speculative {
@@ -865,11 +922,13 @@ struct common_speculative {
     common_speculative_state * curr_impl = nullptr; // current implementation in use (for stats)
 };
 
-static common_ngram_map get_common_ngram_map(const common_speculative_config & config) {
-    uint16_t size_key   = config.params.ngram_size_n;
-    uint16_t size_value = config.params.ngram_size_m;
-    bool     key_only   = (config.type == COMMON_SPECULATIVE_TYPE_NGRAM_MAP_K);
-    uint16_t min_hits   = config.params.ngram_min_hits;
+static common_ngram_map get_common_ngram_map(
+        common_speculative_type type,
+        const common_params_speculative_ngram_map & config) {
+    uint16_t size_key   = config.size_n;
+    uint16_t size_value = config.size_m;
+    bool     key_only   = type == COMMON_SPECULATIVE_TYPE_NGRAM_MAP_K;
+    uint16_t min_hits   = config.min_hits;
 
     return common_ngram_map(size_key, size_value, key_only, min_hits);
 }
@@ -927,8 +986,8 @@ common_speculative * common_speculative_init(
         common_params_speculative & params,
         llama_context             * ctx_tgt) {
     llama_context * ctx_dft = nullptr;
-    if (params.model_dft) {
-        ctx_dft = llama_init_from_model(params.model_dft, params.cparams_dft);
+    if (params.draft.model) {
+        ctx_dft = llama_init_from_model(params.draft.model, params.draft.cparams);
         if (ctx_dft == nullptr) {
             LOG_ERR("%s", "failed to create draft context\n");
             return nullptr;
@@ -938,7 +997,7 @@ common_speculative * common_speculative_init(
     // Compute the implementations to use based on the config and their order of preference
     std::vector<common_speculative_config> configs = {}; // list of speculative configs to try
     {
-        bool has_draft = !params.mparams_dft.path.empty();
+        bool has_draft = !params.draft.mparams.path.empty();
         bool has_draft_eagle3 = false; // TODO PR-18039: if params.speculative.eagle3
 
         bool has_ngram_cache   = (params.type == COMMON_SPECULATIVE_TYPE_NGRAM_CACHE);
@@ -961,16 +1020,17 @@ common_speculative * common_speculative_init(
             configs.push_back(common_speculative_config(COMMON_SPECULATIVE_TYPE_NGRAM_MAP_K4V, params));
         }
         if (has_ngram_mod) {
-            // shared instance for all speculative decoding contexts
-            if (!params.ngram_mod) {
-                params.ngram_mod = std::make_shared<common_ngram_mod>(params.ngram_size_n, 4*1024*1024);
+            auto & sparams = params.ngram_mod;
 
-                LOG_INF("%s: initialized ngram_mod with n=%d, size=%zu (%.3f MB)\n", __func__,
-                        params.ngram_size_n, params.ngram_mod->size(),
-                        (float)(params.ngram_mod->size_bytes())/1024/1024);
+            if (!sparams.obj) {
+                sparams.obj = std::make_shared<common_ngram_mod>(sparams.n_match, 4*1024*1024);
 
-                if (params.ngram_size_n < 16) {
-                    LOG_WRN("%s: ngram_mod n=%d is too small - poor quality is possible, see: https://github.com/ggml-org/llama.cpp/pull/19164\n", __func__, params.ngram_size_n);
+                LOG_INF("%s: initialized ngram_mod with n_match=%d, size=%zu (%.3f MB)\n", __func__,
+                        sparams.n_match, sparams.obj->size(), (float)(sparams.obj->size_bytes())/1024/1024);
+
+                if (sparams.n_match < 16) {
+                    LOG_WRN("%s: ngram_mod n_match=%d is too small - poor quality is possible, "
+                            "see: https://github.com/ggml-org/llama.cpp/pull/19164\n", __func__, sparams.n_match);
                 }
             }
 
@@ -1000,7 +1060,7 @@ common_speculative * common_speculative_init(
                 impls.push_back(std::make_unique<common_speculative_state_draft>(config.type,
                     /* .ctx_tgt      = */ ctx_tgt,
                     /* .ctx_dft      = */ ctx_dft,
-                    /* .replacements = */ params.replacements,
+                    /* .replacements = */ params.draft.replacements,
                     /* .use_ckpt     = */ use_ckpt
                 ));
                 break;
@@ -1010,18 +1070,18 @@ common_speculative * common_speculative_init(
                 break;
             }
             case COMMON_SPECULATIVE_TYPE_NGRAM_SIMPLE: {
-                common_ngram_map ngram_map = get_common_ngram_map(config);
+                common_ngram_map ngram_map = get_common_ngram_map(config.type, config.params.ngram_simple);
 
                 uint16_t ngram_size_key   = ngram_map.size_key;
                 uint16_t mgram_size_value = ngram_map.size_value;
 
                 auto config_simple = common_ngram_simple_config {
-                    /* .size_ngram      = */ ngram_size_key,
-                    /* .size_mgram      = */ mgram_size_value
+                    /* .size_ngram = */ ngram_size_key,
+                    /* .size_mgram = */ mgram_size_value
                 };
                 auto state = std::make_unique<common_speculative_state_ngram_simple>(
-                    /* .type            = */ config.type,
-                    /* .state           = */ config_simple
+                    /* .type  = */ config.type,
+                    /* .state = */ config_simple
                 );
                 impls.push_back(std::move(state));
                 break;
@@ -1030,18 +1090,17 @@ common_speculative * common_speculative_init(
             case COMMON_SPECULATIVE_TYPE_NGRAM_MAP_K4V: {
                 impls.push_back(std::make_unique<common_speculative_state_ngram_map_k>(
                     (config.type),
-                    get_common_ngram_map(config)
+                    get_common_ngram_map(config.type, config.params.ngram_map_k)
                 ));
                 break;
             }
             case COMMON_SPECULATIVE_TYPE_NGRAM_MOD: {
-                GGML_ASSERT(config.params.ngram_mod);
-                impls.push_back(std::make_unique<common_speculative_state_ngram_mod>(config.type, *config.params.ngram_mod));
+                GGML_ASSERT(config.params.ngram_mod.obj);
+                impls.push_back(std::make_unique<common_speculative_state_ngram_mod>(config.type, *config.params.ngram_mod.obj));
                 break;
             }
             case COMMON_SPECULATIVE_TYPE_NGRAM_CACHE: {
-                auto state = create_state_ngram_cache(
-                        params.lookup_cache_static, params.lookup_cache_dynamic, config);
+                auto state = create_state_ngram_cache(params.ngram_cache.lookup_cache_static, params.ngram_cache.lookup_cache_dynamic, config);
                 impls.push_back(std::make_unique<common_speculative_state_ngram_cache>(state));
                 break;
             }
@@ -1099,6 +1158,15 @@ llama_tokens common_speculative_draft(
             impl->n_call_draft++;
         }
 
+        {
+            const int n_min = impl->n_min(params);
+
+            if (!result.empty() && (int) result.size() < n_min) {
+                LOG_DBG("%s: ignoring small draft: %d < %d\n", __func__, (int) result.size(), n_min);
+                result.clear();
+            }
+        }
+
         if (!result.empty()) {
             LOG_DBG("%s: called impl %s, hist size = %zu, call_count = %zu, gen = %zu\n", __func__,
                     common_speculative_type_to_str(impl.get()->type).c_str(), prompt_tgt.size(),
@@ -1108,7 +1176,7 @@ llama_tokens common_speculative_draft(
             impl->n_gen_drafts++;
             impl->n_gen_tokens += result.size();
 
-            break; // We have a draft, so break out of the loop and return it.
+            break; // we have a draft, so break out of the loop and return it.
         }
     }
 
@@ -1136,6 +1204,32 @@ void common_speculative_accept(common_speculative * spec, uint16_t n_accepted) {
     }
 }
 
+int32_t common_speculative_n_max(const common_speculative * spec, const common_params_speculative & params) {
+    if (spec == nullptr) {
+        return 0;
+    }
+
+    int32_t n_max = 0;
+    for (const auto & impl : spec->impls) {
+        n_max = std::max(n_max, impl->n_max(params));
+    }
+
+    return n_max;
+}
+
+int32_t common_speculative_n_min(const common_speculative * spec, const common_params_speculative & params) {
+    if (spec == nullptr) {
+        return 0;
+    }
+
+    int32_t n_min = 0;
+    for (const auto & impl : spec->impls) {
+        n_min = std::max(n_min, impl->n_min(params));
+    }
+
+    return n_min;
+}
+
 void common_speculative_print_stats(const common_speculative * spec) {
     if (spec == nullptr) {
         return;

common/speculative.h

@@ -33,6 +33,9 @@ llama_tokens common_speculative_draft(
 // informs the speculative decoder that n_accepted tokens were accepted by the target model
 void common_speculative_accept(common_speculative * spec, uint16_t n_accepted);
 
+int32_t common_speculative_n_max(const common_speculative * spec, const common_params_speculative & params);
+int32_t common_speculative_n_min(const common_speculative * spec, const common_params_speculative & params);
+
 // print statistics about the speculative decoding
 void common_speculative_print_stats(const common_speculative * spec);
 

ggml/src/ggml-backend-reg.cpp

@@ -181,6 +181,12 @@ struct ggml_backend_registry {
             return;
         }
 
+        for (auto & entry : backends) {
+            if (entry.reg == reg) {
+                return;
+            }
+        }
+
 #ifndef NDEBUG
         GGML_LOG_DEBUG("%s: registered backend %s (%zu devices)\n",
             __func__, ggml_backend_reg_name(reg), ggml_backend_reg_dev_count(reg));
@@ -192,6 +198,12 @@ struct ggml_backend_registry {
     }
 
     void register_device(ggml_backend_dev_t device) {
+        for (auto & dev : devices) {
+            if (dev == device) {
+                return;
+            }
+        }
+
 #ifndef NDEBUG
         GGML_LOG_DEBUG("%s: registered device %s (%s)\n", __func__, ggml_backend_dev_name(device), ggml_backend_dev_description(device));
 #endif

ggml/src/ggml-vulkan/ggml-vulkan.cpp

@@ -13041,6 +13041,7 @@ static bool ggml_vk_build_graph(ggml_backend_vk_context * ctx, ggml_cgraph * cgr
             if (vk_perf_logger_enabled && vk_perf_logger_concurrent) {
                 ctx->query_node_idx[ctx->query_idx] = node_idx;
                 compute_ctx->s->buffer->buf.writeTimestamp(vk::PipelineStageFlagBits::eAllCommands, ctx->query_pool, ctx->query_idx++);
+                ggml_vk_sync_buffers(ctx, compute_ctx);
             }
         }
         // Add all fused nodes to the unsynchronized lists.
@@ -14530,6 +14531,7 @@ static ggml_status ggml_backend_vk_graph_compute(ggml_backend_t backend, ggml_cg
         compute_ctx = ggml_vk_get_compute_ctx(ctx);
         ctx->query_idx = 0;
         compute_ctx->s->buffer->buf.writeTimestamp(vk::PipelineStageFlagBits::eAllCommands, ctx->query_pool, ctx->query_idx++);
+        ggml_vk_sync_buffers(ctx, compute_ctx);
     }
 
     ctx->prealloc_y_last_pipeline_used = nullptr;
@@ -14766,6 +14768,7 @@ static ggml_status ggml_backend_vk_graph_compute(ggml_backend_t backend, ggml_cg
                 ctx->query_nodes[ctx->query_idx] = cgraph->nodes[i];
                 ctx->query_fusion_names[ctx->query_idx] = fusion_string;
                 compute_ctx->s->buffer->buf.writeTimestamp(vk::PipelineStageFlagBits::eAllCommands, ctx->query_pool, ctx->query_idx++);
+                ggml_vk_sync_buffers(ctx, compute_ctx);
             } else {
                 // track a fusion string and number of fused ops for the current node_idx
                 ctx->query_fusion_names[i] = fusion_string;

ggml/src/ggml-vulkan/vulkan-shaders/mul_mat_vecq_funcs.glsl

@@ -296,13 +296,22 @@ vec2 get_dm_scale(uint ib, uint iqs) {
     const uint ib_k = ib / 8;
     const uint iqs_k = (ib % 8) * 8 + iqs;
     const uint is = iqs_k / 8;
-    u8vec2 scale_dm;
-    if (is < 4) {
-        scale_dm = u8vec2(data_a[ib_k].scales[is] & 0x3F, data_a[ib_k].scales[is + 4] & 0x3F);
-    } else {
-        scale_dm = u8vec2((data_a[ib_k].scales[is+4] & 0xF) | ((data_a[ib_k].scales[is-4] & 0xC0) >> 2),
-                          (data_a[ib_k].scales[is+4] >>  4) | ((data_a[ib_k].scales[is  ] & 0xC0) >> 2));
-    }
+
+    const uvec3 scales = uvec3(data_a_packed32[ib_k].scales[0],
+                               data_a_packed32[ib_k].scales[1],
+                               data_a_packed32[ib_k].scales[2]);
+    const uint scalesoffs = (is & 3) * 8;
+
+    const uint scidx0 = (is < 4) ? 0 : 2;
+    const uint scidxshift0 = scalesoffs;
+    const uint scidxshift1 = (is < 4) ? scalesoffs : scalesoffs + 2;
+    const uint mbidx0 = (is < 4) ? 1 : 2;
+    const uint mbidxshift0 = (is < 4) ? scalesoffs : scalesoffs + 4;
+    const uint mbidxshift1 = (is < 4) ? scalesoffs : scalesoffs + 2;
+
+    const uint8_t sc    = uint8_t(((scales[scidx0] >> scidxshift0) & 0xF) | ((scales[0] >> scidxshift1) & 0x30));
+    const uint8_t mbyte = uint8_t(((scales[mbidx0] >> mbidxshift0) & 0xF) | ((scales[1] >> mbidxshift1) & 0x30));
+    u8vec2 scale_dm = u8vec2(sc, mbyte);
 
     return FLOAT_TYPEV2(data_a_packed32[ib_k].dm) * FLOAT_TYPEV2(scale_dm);
 }

ggml/src/ggml-vulkan/vulkan-shaders/mul_mm_funcs.glsl

@@ -201,19 +201,20 @@ void load_a_to_shmem(const uint pos_a, const uint row, const uint col, const uin
 
             const vec2 loadd = vec2(data_a[ib].dm);
 
-            const uint scidx0 = (is < 4) ? is : (is + 4);
-            const uint scidx1 = (is < 4) ? is : (is - 4);
-            const uint scidxmask1 = (is < 4) ? 0x30 : 0xC0;
-            const uint scidxshift1 = (is < 4) ? 0 : 2;
-            const uint mbidx0 = is + 4;
-            const uint mbidx1 = (is < 4) ? is + 4 : is;
-            const uint mbidxmask0 = (is < 4) ? 0xF : 0xF0;
-            const uint mbidxshift0 = (is < 4) ? 0 : 4;
-            const uint mbidxmask1 = (is < 4) ? 0x30 : 0xC0;
-            const uint mbidxshift1 = (is < 4) ? 0 : 2;
+            const uvec3 scales = uvec3(data_a_packed32[ib].scales[0],
+                                       data_a_packed32[ib].scales[1],
+                                       data_a_packed32[ib].scales[2]);
+            const uint scalesoffs = (is & 3) * 8;
 
-            const uint8_t sc = uint8_t((data_a[ib].scales[scidx0] & 0xF) | ((data_a[ib].scales[scidx1] & scidxmask1) >> scidxshift1));
-            const uint8_t mbyte = uint8_t((data_a[ib].scales[mbidx0] & mbidxmask0) >> mbidxshift0 | ((data_a[ib].scales[mbidx1] & mbidxmask1) >> mbidxshift1));
+            const uint scidx0 = (is < 4) ? 0 : 2;
+            const uint scidxshift0 = scalesoffs;
+            const uint scidxshift1 = (is < 4) ? scalesoffs : scalesoffs + 2;
+            const uint mbidx0 = (is < 4) ? 1 : 2;
+            const uint mbidxshift0 = (is < 4) ? scalesoffs : scalesoffs + 4;
+            const uint mbidxshift1 = (is < 4) ? scalesoffs : scalesoffs + 2;
+
+            const uint8_t sc    = uint8_t(((scales[scidx0] >> scidxshift0) & 0xF) | ((scales[0] >> scidxshift1) & 0x30));
+            const uint8_t mbyte = uint8_t(((scales[mbidx0] >> mbidxshift0) & 0xF) | ((scales[1] >> mbidxshift1) & 0x30));
 
             const float d = loadd.x * sc;
             const float m = -loadd.y * mbyte;
@@ -237,19 +238,20 @@ void load_a_to_shmem(const uint pos_a, const uint row, const uint col, const uin
 
             const vec2 loadd = vec2(data_a[ib].dm);
 
-            const uint scidx0 = (is < 4) ? is : (is + 4);
-            const uint scidx1 = (is < 4) ? is : (is - 4);
-            const uint scidxmask1 = (is < 4) ? 0x30 : 0xC0;
-            const uint scidxshift1 = (is < 4) ? 0 : 2;
-            const uint mbidx0 = is + 4;
-            const uint mbidx1 = (is < 4) ? is + 4 : is;
-            const uint mbidxmask0 = (is < 4) ? 0xF : 0xF0;
-            const uint mbidxshift0 = (is < 4) ? 0 : 4;
-            const uint mbidxmask1 = (is < 4) ? 0x30 : 0xC0;
-            const uint mbidxshift1 = (is < 4) ? 0 : 2;
+            const uvec3 scales = uvec3(data_a_packed32[ib].scales[0],
+                                       data_a_packed32[ib].scales[1],
+                                       data_a_packed32[ib].scales[2]);
+            const uint scalesoffs = (is & 3) * 8;
 
-            const uint8_t sc    = uint8_t((data_a[ib].scales[scidx0] & 0xF)                         | ((data_a[ib].scales[scidx1] & scidxmask1) >> scidxshift1));
-            const uint8_t mbyte = uint8_t(((data_a[ib].scales[mbidx0] & mbidxmask0) >> mbidxshift0) | ((data_a[ib].scales[mbidx1] & mbidxmask1) >> mbidxshift1));
+            const uint scidx0 = (is < 4) ? 0 : 2;
+            const uint scidxshift0 = scalesoffs;
+            const uint scidxshift1 = (is < 4) ? scalesoffs : scalesoffs + 2;
+            const uint mbidx0 = (is < 4) ? 1 : 2;
+            const uint mbidxshift0 = (is < 4) ? scalesoffs : scalesoffs + 4;
+            const uint mbidxshift1 = (is < 4) ? scalesoffs : scalesoffs + 2;
+
+            const uint8_t sc    = uint8_t(((scales[scidx0] >> scidxshift0) & 0xF) | ((scales[0] >> scidxshift1) & 0x30));
+            const uint8_t mbyte = uint8_t(((scales[mbidx0] >> mbidxshift0) & 0xF) | ((scales[1] >> mbidxshift1) & 0x30));
 
             const float d = loadd.x * sc;
             const float m = -loadd.y * mbyte;

tools/server/server-chat.cpp

@@ -575,14 +575,14 @@ json server_chat_msg_diff_to_json_oaicompat(const common_chat_msg_diff & diff) {
 json convert_transcriptions_to_chatcmpl(
         const json & inp_body,
         const common_chat_templates * tmpls,
-        const std::map<std::string, raw_buffer> & in_files,
+        const std::map<std::string, uploaded_file> & in_files,
         std::vector<raw_buffer> & out_files) {
     // TODO @ngxson : this function may need to be improved in the future
     // handle input files
     out_files.clear();
     auto it = in_files.find("file");
     if (it != in_files.end()) {
-        out_files.push_back(it->second);
+        out_files.push_back(it->second.data);
     } else {
         throw std::invalid_argument("No input file found for transcription");
     }

tools/server/server-chat.h

@@ -4,6 +4,7 @@
 
 #include "chat.h"
 #include "server-common.h"
+#include "server-http.h"
 
 #include <nlohmann/json_fwd.hpp>
 
@@ -19,7 +20,7 @@ json server_chat_convert_anthropic_to_oai(const json & body);
 json convert_transcriptions_to_chatcmpl(
     const json & body,
     const common_chat_templates * tmpls,
-    const std::map<std::string, raw_buffer> & in_files,
+    const std::map<std::string, uploaded_file> & in_files,
     std::vector<raw_buffer> & out_files);
 
 json server_chat_msg_diff_to_json_oaicompat(const common_chat_msg_diff & diff);

tools/server/server-context.cpp

@@ -309,8 +309,10 @@ struct server_slot {
             return 0;
         }
 
+        const int n_draft_min = common_speculative_n_min(spec.get(), task->params.speculative);
+
         // determine the max draft that fits the current slot state
-        int n_draft_max = task->params.speculative.n_max;
+        int n_draft_max = common_speculative_n_max(spec.get(), task->params.speculative);
 
         // note: slot.prompt is not yet expanded with the `id` token sampled above
         //       also, need to leave space for 1 extra token to allow context shifts
@@ -322,8 +324,8 @@ struct server_slot {
 
         SLT_DBG(*this, "max possible draft: %d\n", n_draft_max);
 
-        if (n_draft_max < task->params.speculative.n_min) {
-            SLT_DBG(*this, "the max possible draft is too small: %d < %d - skipping speculative decoding\n", n_draft_max, task->params.speculative.n_min);
+        if (n_draft_max < n_draft_min) {
+            SLT_DBG(*this, "the max possible draft is too small: %d < %d - skipping speculative decoding\n", n_draft_max, n_draft_min);
             n_draft_max = 0;
         }
 
@@ -358,11 +360,6 @@ struct server_slot {
                     spec_draft.resize(n_draft_max);
                 }
 
-                if (spec_draft.size() < (size_t) params_spec.n_min) {
-                    SLT_DBG(*this, "ignoring small draft: %d < %d\n", (int) spec_draft.size(), params_spec.n_min);
-                    spec_draft.clear();
-                }
-
                 if (!spec_draft.empty() && ctx_seq_rm_type == COMMON_CONTEXT_SEQ_RM_TYPE_FULL) {
                     const auto n_tokens = prompt.tokens.size();
 
@@ -770,9 +767,9 @@ private:
 
         if (params_base.speculative.has_dft()) {
             // TODO speculative: move to common/speculative.cpp?
-            SRV_INF("loading draft model '%s'\n", params_base.speculative.mparams_dft.path.c_str());
+            const auto & params_spec = params_base.speculative.draft;
 
-            const auto & params_spec = params_base.speculative;
+            SRV_INF("loading draft model '%s'\n", params_spec.mparams.path.c_str());
 
             auto params_dft = params_base;
 
@@ -780,7 +777,7 @@ private:
             params_dft.n_ctx        = params_spec.n_ctx == 0 ? llama_n_ctx_seq(ctx) : params_spec.n_ctx;
             params_dft.n_batch      = llama_n_ctx_seq(ctx);
             params_dft.devices      = params_spec.devices;
-            params_dft.model        = params_spec.mparams_dft;
+            params_dft.model        = params_spec.mparams;
             params_dft.n_gpu_layers = params_spec.n_gpu_layers;
             params_dft.cache_type_k = params_spec.cache_type_k;
             params_dft.cache_type_v = params_spec.cache_type_v;
@@ -800,8 +797,8 @@ private:
                 return false;
             }
 
-            params_base.speculative.model_dft = model_dft.get();
-            params_base.speculative.cparams_dft = common_context_params_to_llama(params_dft);
+            params_base.speculative.draft.model = model_dft.get();
+            params_base.speculative.draft.cparams = common_context_params_to_llama(params_dft);
         }
 
         std::string & mmproj_path = params_base.mmproj.path;
@@ -1310,7 +1307,7 @@ private:
             backend_sampling &= task.params.sampling.backend_sampling;
 
             // TODO: speculative decoding requires multiple samples per batch - not supported yet
-            backend_sampling &= !(slot.can_speculate() && task.params.speculative.n_max > 0);
+            backend_sampling &= !(slot.can_speculate() && common_speculative_n_max(slot.spec.get(), task.params.speculative) > 0);
 
             // TODO: getting post/pre sampling logits is not yet supported with backend sampling
             backend_sampling &= !need_logits;
@@ -3031,7 +3028,7 @@ private:
                 slot.sampled = ids.back(); // last accepted token
                 SLT_DBG(slot, "add accepted tokens: sampled=%d, ids.size=%zu, n_draft=%zu\n", slot.sampled, ids.size(), n_draft);
 
-                llama_memory_seq_rm(llama_get_memory(slot.ctx), slot.id, slot.prompt.n_tokens(), -1);
+                llama_memory_seq_rm(llama_get_memory(slot.ctx), slot.id, slot.prompt.tokens.pos_next(), -1);
 
                 for (size_t i = 0; i < ids.size(); ++i) {
                     completion_token_output result;

tools/server/server-cors-proxy.h

@@ -49,6 +49,7 @@ static server_http_res_ptr proxy_request(const server_http_req & req, std::strin
             parsed_url.path,
             headers,
             req.body,
+            req.files,
             req.should_stop,
             600, // timeout_read (default to 10 minutes)
             600  // timeout_write (default to 10 minutes)

tools/server/server-http.cpp

@@ -438,7 +438,7 @@ void server_http_context::get(const std::string & path, const server_http_contex
 void server_http_context::post(const std::string & path, const server_http_context::handler_t & handler) const {
     pimpl->srv->Post(path_prefix + path, [handler](const httplib::Request & req, httplib::Response & res) {
         std::string body = req.body;
-        std::map<std::string, raw_buffer> files;
+        std::map<std::string, uploaded_file> files;
 
         if (req.is_multipart_form_data()) {
             // translate text fields to a JSON object and use it as the body
@@ -459,7 +459,11 @@ void server_http_context::post(const std::string & path, const server_http_conte
 
             // populate files from multipart form
             for (const auto & [key, file] : req.form.files) {
-                files[key] = raw_buffer(file.content.begin(), file.content.end());
+                files[key] = uploaded_file{
+                    raw_buffer(file.content.begin(), file.content.end()),
+                    file.filename,
+                    file.content_type,
+                };
             }
         }
 

tools/server/server-http.h

@@ -36,13 +36,19 @@ struct server_http_res {
 using server_http_res_ptr = std::unique_ptr<server_http_res>;
 using raw_buffer = std::vector<uint8_t>;
 
+struct uploaded_file {
+    raw_buffer data;
+    std::string filename;
+    std::string content_type;
+};
+
 struct server_http_req {
     std::map<std::string, std::string> params; // path_params + query_params
     std::map<std::string, std::string> headers; // used by MCP proxy
     std::string path;
     std::string query_string; // query parameters string (e.g. "action=save")
     std::string body;
-    std::map<std::string, raw_buffer> files; // used for file uploads (form data)
+    std::map<std::string, uploaded_file> files; // used for file uploads (form data)
     const std::function<bool()> & should_stop;
 
     std::string get_param(const std::string & key, const std::string & def = "") const {

tools/server/server-models.cpp

@@ -18,6 +18,8 @@
 #include <chrono>
 #include <queue>
 #include <filesystem>
+#include <random>
+#include <sstream>
 #include <cstring>
 
 #ifdef _WIN32
@@ -823,6 +825,7 @@ server_http_res_ptr server_models::proxy_request(const server_http_req & req, co
             proxy_path,
             req.headers,
             req.body,
+            req.files,
             req.should_stop,
             base_params.timeout_read,
             base_params.timeout_write
@@ -1126,6 +1129,77 @@ static bool should_strip_proxy_header(const std::string & header_name) {
     return false;
 }
 
+static std::string generate_multipart_boundary() {
+    thread_local std::mt19937 gen(std::random_device{}());
+    static const char chars[] = "0123456789abcdefghijklmnopqrstuvwxyz";
+    std::uniform_int_distribution<> dis(0, sizeof(chars) - 2);
+    std::string boundary = "----llama-cpp-proxy-";
+    for (int i = 0; i < 16; i++) {
+        boundary += chars[dis(gen)];
+    }
+    return boundary;
+}
+
+static std::string build_multipart_body(
+        const json & form_fields,
+        const std::map<std::string, uploaded_file> & files,
+        const std::string & boundary) {
+    static auto sanitize_field = [](const std::string & text) {
+        std::string result;
+        result.reserve(text.size());
+        for (char c : text) {
+            if (c != '\n' && c != '\r' && c != '"') {
+                result += c;
+            }
+        }
+        return result;
+    };
+
+    std::ostringstream body;
+
+    for (const auto & [key, value] : form_fields.items()) {
+        if (value.is_array()) {
+            for (const auto & item : value) {
+                body << "--" << boundary << "\r\n";
+                body << "Content-Disposition: form-data; name=\"" << sanitize_field(key) << "\"\r\n";
+                body << "\r\n";
+                if (!item.is_string()) {
+                    throw std::invalid_argument("expected string");
+                }
+                body << item.get<std::string>() << "\r\n";
+            }
+        } else {
+            body << "--" << boundary << "\r\n";
+            body << "Content-Disposition: form-data; name=\"" << sanitize_field(key) << "\"\r\n";
+            body << "\r\n";
+            if (!value.is_string()) {
+                throw std::invalid_argument("expected string");
+            }
+            body << value.get<std::string>() << "\r\n";
+        }
+    }
+
+    for (const auto & [key, file] : files) {
+        body << "--" << boundary << "\r\n";
+        body << "Content-Disposition: form-data; name=\"" << sanitize_field(key) << "\"";
+        if (!file.filename.empty()) {
+            body << "; filename=\"" << sanitize_field(file.filename) << "\"";
+        }
+        body << "\r\n";
+        if (!file.content_type.empty()) {
+            body << "Content-Type: " << sanitize_field(file.content_type) << "\r\n";
+        } else {
+            body << "Content-Type: application/octet-stream\r\n";
+        }
+        body << "\r\n";
+        body.write(reinterpret_cast<const char*>(file.data.data()), file.data.size());
+        body << "\r\n";
+    }
+
+    body << "--" << boundary << "--\r\n";
+    return body.str();
+}
+
 server_http_proxy::server_http_proxy(
         const std::string & method,
         const std::string & scheme,
@@ -1134,6 +1208,7 @@ server_http_proxy::server_http_proxy(
         const std::string & path,
         const std::map<std::string, std::string> & headers,
         const std::string & body,
+        const std::map<std::string, uploaded_file> & files,
         const std::function<bool()> should_stop,
         int32_t timeout_read,
         int32_t timeout_write
@@ -1195,28 +1270,65 @@ server_http_proxy::server_http_proxy(
         return pipe->write({{}, 0, std::string(data, data_length), ""});
     };
 
+    // when files are present, the body was converted from multipart form data to JSON
+    // we need to reconstruct the multipart body for the downstream server
+    std::string effective_body = body;
+    std::string override_content_type;
+    bool has_files = !files.empty();
+
+    if (has_files) {
+        json form_fields = json::parse(body, nullptr, false);
+        if (!form_fields.is_discarded()) {
+            auto boundary = generate_multipart_boundary();
+            effective_body = build_multipart_body(form_fields, files, boundary);
+            override_content_type = "multipart/form-data; boundary=" + boundary;
+        } else {
+            throw std::runtime_error("failed to parse multipart form fields JSON");
+        }
+    }
+
     // prepare the request to destination server
     httplib::Request req;
     {
         req.method = method;
         req.path = path;
         for (const auto & [key, value] : headers) {
-            if (key == "Accept-Encoding") {
+            const auto lowered = to_lower_copy(key);
+            if (lowered == "accept-encoding") {
                 // disable Accept-Encoding to avoid compressed responses
                 continue;
             }
-            if (key == "Transfer-Encoding") {
+            if (lowered == "transfer-encoding") {
                 // the body is already decoded
                 continue;
             }
-            if (key == "Host" || key == "host") {
+            if (lowered == "content-length") {
+                // let httplib calculate Content-Length from the actual body
+                continue;
+            }
+            if (lowered == "content-type") {
+                if (has_files) {
+                    // we set our own Content-Type with the new boundary
+                    continue;
+                }
+                // when no files but the original request was multipart,
+                // the body is now JSON, so correct the Content-Type
+                if (value.find("multipart/form-data") != std::string::npos) {
+                    override_content_type = "application/json; charset=utf-8";
+                    continue;
+                }
+            }
+            if (lowered == "host") {
                 bool is_default_port = (scheme == "https" && port == 443) || (scheme == "http" && port == 80);
                 req.set_header(key, is_default_port ? host : host + ":" + std::to_string(port));
             } else {
                 req.set_header(key, value);
             }
         }
-        req.body = body;
+        req.body = effective_body;
+        if (!override_content_type.empty()) {
+            req.set_header("Content-Type", override_content_type);
+        }
         req.response_handler = response_handler;
         req.content_receiver = content_receiver;
     }

tools/server/server-models.h

@@ -202,6 +202,7 @@ public:
                       const std::string & path,
                       const std::map<std::string, std::string> & headers,
                       const std::string & body,
+                      const std::map<std::string, uploaded_file> & files,
                       const std::function<bool()> should_stop,
                       int32_t timeout_read,
                       int32_t timeout_write

tools/server/server-task.cpp

@@ -76,13 +76,7 @@ json task_params::to_json(bool only_metrics) const {
             {"reasoning_in_content",      chat_parser_params.reasoning_in_content},
             {"generation_prompt",         chat_parser_params.generation_prompt},
             {"samplers",                  samplers},
-            {"speculative.n_max",         speculative.n_max},
-            {"speculative.n_min",         speculative.n_min},
-            {"speculative.p_min",         speculative.p_min},
             {"speculative.type",          common_speculative_type_to_str(speculative.type)},
-            {"speculative.ngram_size_n",  speculative.ngram_size_n},
-            {"speculative.ngram_size_m",  speculative.ngram_size_m},
-            {"speculative.ngram_m_hits",  speculative.ngram_min_hits},
             {"timings_per_token",         timings_per_token},
             {"post_sampling_probs",       post_sampling_probs},
             {"backend_sampling",          sampling.backend_sampling},
@@ -139,13 +133,7 @@ json task_params::to_json(bool only_metrics) const {
         {"reasoning_in_content",      chat_parser_params.reasoning_in_content},
         {"generation_prompt",         chat_parser_params.generation_prompt},
         {"samplers",                  samplers},
-        {"speculative.n_max",         speculative.n_max},
-        {"speculative.n_min",         speculative.n_min},
-        {"speculative.p_min",         speculative.p_min},
         {"speculative.type",          common_speculative_type_to_str(speculative.type)},
-        {"speculative.ngram_size_n",  speculative.ngram_size_n},
-        {"speculative.ngram_size_m",  speculative.ngram_size_m},
-        {"speculative.ngram_m_hits",  speculative.ngram_min_hits},
         {"timings_per_token",         timings_per_token},
         {"post_sampling_probs",       post_sampling_probs},
         {"backend_sampling",          sampling.backend_sampling},
@@ -308,14 +296,17 @@ task_params server_task::params_from_json_cmpl(
 
     params.speculative = defaults.speculative;
 
-    params.speculative.n_min = json_value(data, "speculative.n_min", defaults.speculative.n_min);
-    params.speculative.n_max = json_value(data, "speculative.n_max", defaults.speculative.n_max);
-    params.speculative.p_min = json_value(data, "speculative.p_min", defaults.speculative.p_min);
+    // TODO: for now, be able to adjust only the draft-model based speculative parameters
+    params.speculative.draft.n_min = json_value(data, "speculative.n_min", defaults.speculative.draft.n_min);
+    params.speculative.draft.n_max = json_value(data, "speculative.n_max", defaults.speculative.draft.n_max);
+    params.speculative.draft.p_min = json_value(data, "speculative.p_min", defaults.speculative.draft.p_min);
 
-    params.speculative.n_min = std::min(params.speculative.n_max, params.speculative.n_min);
-    params.speculative.n_min = std::max(params.speculative.n_min, 0);
-    params.speculative.n_max = std::max(params.speculative.n_max, 0);
+    params.speculative.draft.n_min = std::min(params.speculative.draft.n_max, params.speculative.draft.n_min);
+    params.speculative.draft.n_min = std::max(params.speculative.draft.n_min, 0);
+    params.speculative.draft.n_max = std::max(params.speculative.draft.n_max, 0);
 
+#if 0
+    // for debugging and research purposes
     params.speculative.type = common_speculative_type_from_name(json_value(data, "speculative.type", common_speculative_type_to_str(defaults.speculative.type)));
 
     params.speculative.ngram_size_n     = json_value(data, "speculative.ngram_size_n", defaults.speculative.ngram_size_n);
@@ -325,6 +316,7 @@ task_params server_task::params_from_json_cmpl(
     params.speculative.ngram_size_n     = std::max(std::min(1, (int) params.speculative.ngram_size_n),     1024);
     params.speculative.ngram_size_m     = std::max(std::min(1, (int) params.speculative.ngram_size_m),     1024);
     params.speculative.ngram_min_hits   = std::max(std::min(1, (int) params.speculative.ngram_min_hits),   1024);
+#endif
 
     // Use OpenAI API logprobs only if n_probs wasn't provided
     if (data.contains("logprobs") && params.sampling.n_probs == defaults.sampling.n_probs){

tools/server/tests/utils.py

@@ -83,15 +83,14 @@ class ServerProcess:
     kv_unified: bool | None = False
     server_slots: bool | None = False
     pooling: str | None = None
-    draft: int | None = None
     api_key: str | None = None
     models_dir: str | None = None
     models_max: int | None = None
     no_models_autoload: bool | None = None
     lora_files: List[str] | None = None
     enable_ctx_shift: int | None = False
-    draft_min: int | None = None
-    draft_max: int | None = None
+    spec_draft_n_min: int | None = None
+    spec_draft_n_max: int | None = None
     no_webui: bool | None = None
     jinja: bool | None = None
     reasoning_format: Literal['deepseek', 'none', 'nothink'] | None = None
@@ -165,8 +164,6 @@ class ServerProcess:
             server_args.extend(["--threads", self.n_threads])
         if self.n_gpu_layer:
             server_args.extend(["--n-gpu-layers", self.n_gpu_layer])
-        if self.draft is not None:
-            server_args.extend(["--draft", self.draft])
         if self.server_continuous_batching:
             server_args.append("--cont-batching")
         if self.server_embeddings:
@@ -214,10 +211,10 @@ class ServerProcess:
             server_args.append("--context-shift")
         if self.api_key:
             server_args.extend(["--api-key", self.api_key])
-        if self.draft_max:
-            server_args.extend(["--draft-max", self.draft_max])
-        if self.draft_min:
-            server_args.extend(["--draft-min", self.draft_min])
+        if self.spec_draft_n_max:
+            server_args.extend(["--spec-draft-n-max", self.spec_draft_n_max])
+        if self.spec_draft_n_min:
+            server_args.extend(["--spec-draft-n-min", self.spec_draft_n_min])
         if self.no_webui:
             server_args.append("--no-webui")
         if self.no_models_autoload:

tools/server/webui/src/app.css

@@ -29,7 +29,7 @@
 	--chart-3: oklch(0.398 0.07 227.392);
 	--chart-4: oklch(0.828 0.189 84.429);
 	--chart-5: oklch(0.769 0.188 70.08);
-	--sidebar: oklch(0.987 0 0);
+	--sidebar: oklch(0.985 0 0);
 	--sidebar-foreground: oklch(0.145 0 0);
 	--sidebar-primary: oklch(0.205 0 0);
 	--sidebar-primary-foreground: oklch(0.985 0 0);
@@ -77,7 +77,7 @@
 	--chart-3: oklch(0.769 0.188 70.08);
 	--chart-4: oklch(0.627 0.265 303.9);
 	--chart-5: oklch(0.645 0.246 16.439);
-	--sidebar: oklch(0.19 0 0);
+	--sidebar: oklch(0.2 0 0);
 	--sidebar-foreground: oklch(0.985 0 0);
 	--sidebar-primary: oklch(0.488 0.243 264.376);
 	--sidebar-primary-foreground: oklch(0.985 0 0);

tools/server/webui/src/lib/components/app/actions/ActionIcon.svelte

@@ -1,18 +1,20 @@
 <script lang="ts">
-	import { Button } from '$lib/components/ui/button';
+	import { Button, type ButtonVariant, type ButtonSize } from '$lib/components/ui/button';
 	import * as Tooltip from '$lib/components/ui/tooltip';
 	import type { Component } from 'svelte';
+	import { TooltipSide } from '$lib/enums';
 
 	interface Props {
 		icon: Component;
 		tooltip: string;
-		variant?: 'default' | 'destructive' | 'outline' | 'secondary' | 'ghost' | 'link';
-		size?: 'default' | 'sm' | 'lg' | 'icon';
+		variant?: ButtonVariant;
+		size?: ButtonSize;
 		iconSize?: string;
 		class?: string;
 		disabled?: boolean;
 		onclick: (e?: MouseEvent) => void;
 		'aria-label'?: string;
+		tooltipSide?: TooltipSide;
 	}
 
 	let {
@@ -23,6 +25,7 @@
 		class: className = '',
 		disabled = false,
 		iconSize = 'h-3 w-3',
+		tooltipSide = TooltipSide.TOP,
 		onclick,
 		'aria-label': ariaLabel
 	}: Props = $props();
@@ -35,7 +38,7 @@
 			{size}
 			{disabled}
 			{onclick}
-			class="h-6 w-6 p-0 {className} flex"
+			class="h-6 w-6 p-0 {className} flex hover:bg-transparent data-[state=open]:bg-transparent!"
 			aria-label={ariaLabel || tooltip}
 		>
 			{@const IconComponent = icon}
@@ -44,7 +47,7 @@
 		</Button>
 	</Tooltip.Trigger>
 
-	<Tooltip.Content>
+	<Tooltip.Content side={tooltipSide}>
 		<p>{tooltip}</p>
 	</Tooltip.Content>
 </Tooltip.Root>

tools/server/webui/src/lib/components/app/chat/ChatForm/ChatForm.svelte

@@ -300,7 +300,7 @@
 			if (sendOnEnter || isModifier) {
 				event.preventDefault();
 
-				if (!canSubmit || disabled || isLoading || hasLoadingAttachments) return;
+				if (!canSubmit || disabled || hasLoadingAttachments) return;
 
 				onSubmit?.();
 			}
@@ -555,7 +555,7 @@
 	class="relative {className}"
 	onsubmit={(e) => {
 		e.preventDefault();
-		if (!canSubmit || disabled || isLoading || hasLoadingAttachments) return;
+		if (!canSubmit || disabled || hasLoadingAttachments) return;
 		onSubmit?.();
 	}}
 >

tools/server/webui/src/lib/components/app/chat/ChatForm/ChatFormActions/ChatFormActionAttachmentsDropdown.svelte

@@ -1,333 +0,0 @@
-<script lang="ts">
-	import { page } from '$app/state';
-	import { Plus, MessageSquare, Settings, Zap, FolderOpen } from '@lucide/svelte';
-	import { Button } from '$lib/components/ui/button';
-	import * as DropdownMenu from '$lib/components/ui/dropdown-menu';
-	import * as Tooltip from '$lib/components/ui/tooltip';
-	import { Switch } from '$lib/components/ui/switch';
-	import { FILE_TYPE_ICONS, TOOLTIP_DELAY_DURATION } from '$lib/constants';
-	import { McpLogo, DropdownMenuSearchable } from '$lib/components/app';
-	import { conversationsStore } from '$lib/stores/conversations.svelte';
-	import { mcpStore } from '$lib/stores/mcp.svelte';
-
-	import { HealthCheckStatus } from '$lib/enums';
-	import type { MCPServerSettingsEntry } from '$lib/types';
-
-	interface Props {
-		class?: string;
-		disabled?: boolean;
-		hasAudioModality?: boolean;
-		hasVisionModality?: boolean;
-		hasMcpPromptsSupport?: boolean;
-		hasMcpResourcesSupport?: boolean;
-		onFileUpload?: () => void;
-		onSystemPromptClick?: () => void;
-		onMcpPromptClick?: () => void;
-		onMcpSettingsClick?: () => void;
-		onMcpResourcesClick?: () => void;
-	}
-
-	let {
-		class: className = '',
-		disabled = false,
-		hasAudioModality = false,
-		hasVisionModality = false,
-		hasMcpPromptsSupport = false,
-		hasMcpResourcesSupport = false,
-		onFileUpload,
-		onSystemPromptClick,
-		onMcpPromptClick,
-		onMcpSettingsClick,
-		onMcpResourcesClick
-	}: Props = $props();
-
-	let isNewChat = $derived(!page.params.id);
-
-	let systemMessageTooltip = $derived(
-		isNewChat
-			? 'Add custom system message for a new conversation'
-			: 'Inject custom system message at the beginning of the conversation'
-	);
-
-	let dropdownOpen = $state(false);
-
-	let mcpServers = $derived(mcpStore.getServersSorted().filter((s) => s.enabled));
-	let hasMcpServers = $derived(mcpServers.length > 0);
-	let mcpSearchQuery = $state('');
-	let filteredMcpServers = $derived.by(() => {
-		const query = mcpSearchQuery.toLowerCase().trim();
-		if (!query) return mcpServers;
-		return mcpServers.filter((s) => {
-			const name = getServerLabel(s).toLowerCase();
-			const url = s.url.toLowerCase();
-			return name.includes(query) || url.includes(query);
-		});
-	});
-
-	function getServerLabel(server: MCPServerSettingsEntry): string {
-		return mcpStore.getServerLabel(server);
-	}
-
-	function isServerEnabledForChat(serverId: string): boolean {
-		return conversationsStore.isMcpServerEnabledForChat(serverId);
-	}
-
-	async function toggleServerForChat(serverId: string) {
-		await conversationsStore.toggleMcpServerForChat(serverId);
-	}
-
-	function handleMcpSubMenuOpen(open: boolean) {
-		if (open) {
-			mcpSearchQuery = '';
-			mcpStore.runHealthChecksForServers(mcpServers);
-		}
-	}
-
-	function handleMcpPromptClick() {
-		dropdownOpen = false;
-		onMcpPromptClick?.();
-	}
-
-	function handleMcpSettingsClick() {
-		dropdownOpen = false;
-		onMcpSettingsClick?.();
-	}
-
-	function handleMcpResourcesClick() {
-		dropdownOpen = false;
-		onMcpResourcesClick?.();
-	}
-
-	const fileUploadTooltipText = 'Add files, system prompt or MCP Servers';
-</script>
-
-<div class="flex items-center gap-1 {className}">
-	<DropdownMenu.Root bind:open={dropdownOpen}>
-		<DropdownMenu.Trigger name="Attach files" {disabled}>
-			<Tooltip.Root>
-				<Tooltip.Trigger class="w-full">
-					<Button
-						class="file-upload-button h-8 w-8 rounded-full p-0"
-						{disabled}
-						variant="secondary"
-						type="button"
-					>
-						<span class="sr-only">{fileUploadTooltipText}</span>
-
-						<Plus class="h-4 w-4" />
-					</Button>
-				</Tooltip.Trigger>
-
-				<Tooltip.Content>
-					<p>{fileUploadTooltipText}</p>
-				</Tooltip.Content>
-			</Tooltip.Root>
-		</DropdownMenu.Trigger>
-
-		<DropdownMenu.Content align="start" class="w-48">
-			{#if hasVisionModality}
-				<DropdownMenu.Item
-					class="images-button flex cursor-pointer items-center gap-2"
-					onclick={() => onFileUpload?.()}
-				>
-					<FILE_TYPE_ICONS.image class="h-4 w-4" />
-
-					<span>Images</span>
-				</DropdownMenu.Item>
-			{:else}
-				<Tooltip.Root delayDuration={TOOLTIP_DELAY_DURATION}>
-					<Tooltip.Trigger class="w-full">
-						<DropdownMenu.Item
-							class="images-button flex cursor-pointer items-center gap-2"
-							disabled
-						>
-							<FILE_TYPE_ICONS.image class="h-4 w-4" />
-
-							<span>Images</span>
-						</DropdownMenu.Item>
-					</Tooltip.Trigger>
-
-					<Tooltip.Content side="right">
-						<p>Image processing requires a vision model</p>
-					</Tooltip.Content>
-				</Tooltip.Root>
-			{/if}
-
-			{#if hasAudioModality}
-				<DropdownMenu.Item
-					class="audio-button flex cursor-pointer items-center gap-2"
-					onclick={() => onFileUpload?.()}
-				>
-					<FILE_TYPE_ICONS.audio class="h-4 w-4" />
-
-					<span>Audio Files</span>
-				</DropdownMenu.Item>
-			{:else}
-				<Tooltip.Root delayDuration={TOOLTIP_DELAY_DURATION}>
-					<Tooltip.Trigger class="w-full">
-						<DropdownMenu.Item class="audio-button flex cursor-pointer items-center gap-2" disabled>
-							<FILE_TYPE_ICONS.audio class="h-4 w-4" />
-
-							<span>Audio Files</span>
-						</DropdownMenu.Item>
-					</Tooltip.Trigger>
-
-					<Tooltip.Content side="right">
-						<p>Audio files processing requires an audio model</p>
-					</Tooltip.Content>
-				</Tooltip.Root>
-			{/if}
-
-			<DropdownMenu.Item
-				class="flex cursor-pointer items-center gap-2"
-				onclick={() => onFileUpload?.()}
-			>
-				<FILE_TYPE_ICONS.text class="h-4 w-4" />
-
-				<span>Text Files</span>
-			</DropdownMenu.Item>
-
-			{#if hasVisionModality}
-				<DropdownMenu.Item
-					class="flex cursor-pointer items-center gap-2"
-					onclick={() => onFileUpload?.()}
-				>
-					<FILE_TYPE_ICONS.pdf class="h-4 w-4" />
-
-					<span>PDF Files</span>
-				</DropdownMenu.Item>
-			{:else}
-				<Tooltip.Root delayDuration={TOOLTIP_DELAY_DURATION}>
-					<Tooltip.Trigger class="w-full">
-						<DropdownMenu.Item
-							class="flex cursor-pointer items-center gap-2"
-							onclick={() => onFileUpload?.()}
-						>
-							<FILE_TYPE_ICONS.pdf class="h-4 w-4" />
-
-							<span>PDF Files</span>
-						</DropdownMenu.Item>
-					</Tooltip.Trigger>
-
-					<Tooltip.Content side="right">
-						<p>PDFs will be converted to text. Image-based PDFs may not work properly.</p>
-					</Tooltip.Content>
-				</Tooltip.Root>
-			{/if}
-
-			<Tooltip.Root delayDuration={TOOLTIP_DELAY_DURATION}>
-				<Tooltip.Trigger class="w-full">
-					<DropdownMenu.Item
-						class="flex cursor-pointer items-center gap-2"
-						onclick={() => onSystemPromptClick?.()}
-					>
-						<MessageSquare class="h-4 w-4" />
-
-						<span>System Message</span>
-					</DropdownMenu.Item>
-				</Tooltip.Trigger>
-
-				<Tooltip.Content side="right">
-					<p>{systemMessageTooltip}</p>
-				</Tooltip.Content>
-			</Tooltip.Root>
-
-			<DropdownMenu.Separator />
-
-			<DropdownMenu.Sub onOpenChange={handleMcpSubMenuOpen}>
-				<DropdownMenu.SubTrigger class="flex cursor-pointer items-center gap-2">
-					<McpLogo class="h-4 w-4" />
-
-					<span>MCP Servers</span>
-				</DropdownMenu.SubTrigger>
-
-				<DropdownMenu.SubContent class="w-72 pt-0">
-					<DropdownMenuSearchable
-						placeholder="Search servers..."
-						bind:searchValue={mcpSearchQuery}
-						emptyMessage={hasMcpServers ? 'No servers found' : 'No MCP servers configured'}
-						isEmpty={filteredMcpServers.length === 0}
-					>
-						<div class="max-h-64 overflow-y-auto">
-							{#each filteredMcpServers as server (server.id)}
-								{@const healthState = mcpStore.getHealthCheckState(server.id)}
-								{@const hasError = healthState.status === HealthCheckStatus.ERROR}
-								{@const isEnabledForChat = isServerEnabledForChat(server.id)}
-
-								<button
-									type="button"
-									class="flex w-full items-center justify-between gap-2 rounded-sm px-2 py-2 text-left transition-colors hover:bg-accent disabled:cursor-not-allowed disabled:opacity-50"
-									onclick={() => !hasError && toggleServerForChat(server.id)}
-									disabled={hasError}
-								>
-									<div class="flex min-w-0 flex-1 items-center gap-2">
-										{#if mcpStore.getServerFavicon(server.id)}
-											<img
-												src={mcpStore.getServerFavicon(server.id)}
-												alt=""
-												class="h-4 w-4 shrink-0 rounded-sm"
-												onerror={(e) => {
-													(e.currentTarget as HTMLImageElement).style.display = 'none';
-												}}
-											/>
-										{/if}
-
-										<span class="truncate text-sm">{getServerLabel(server)}</span>
-
-										{#if hasError}
-											<span
-												class="shrink-0 rounded bg-destructive/15 px-1.5 py-0.5 text-xs text-destructive"
-											>
-												Error
-											</span>
-										{/if}
-									</div>
-
-									<Switch
-										checked={isEnabledForChat}
-										disabled={hasError}
-										onclick={(e: MouseEvent) => e.stopPropagation()}
-										onCheckedChange={() => toggleServerForChat(server.id)}
-									/>
-								</button>
-							{/each}
-						</div>
-
-						{#snippet footer()}
-							<DropdownMenu.Item
-								class="flex cursor-pointer items-center gap-2"
-								onclick={handleMcpSettingsClick}
-							>
-								<Settings class="h-4 w-4" />
-
-								<span>Manage MCP Servers</span>
-							</DropdownMenu.Item>
-						{/snippet}
-					</DropdownMenuSearchable>
-				</DropdownMenu.SubContent>
-			</DropdownMenu.Sub>
-
-			{#if hasMcpPromptsSupport}
-				<DropdownMenu.Item
-					class="flex cursor-pointer items-center gap-2"
-					onclick={handleMcpPromptClick}
-				>
-					<Zap class="h-4 w-4" />
-
-					<span>MCP Prompt</span>
-				</DropdownMenu.Item>
-			{/if}
-
-			{#if hasMcpResourcesSupport}
-				<DropdownMenu.Item
-					class="flex cursor-pointer items-center gap-2"
-					onclick={handleMcpResourcesClick}
-				>
-					<FolderOpen class="h-4 w-4" />
-
-					<span>MCP Resources</span>
-				</DropdownMenu.Item>
-			{/if}
-		</DropdownMenu.Content>
-	</DropdownMenu.Root>
-</div>

tools/server/webui/src/lib/components/app/chat/ChatForm/ChatFormActions/ChatFormActionAttachmentsSheet.svelte

@@ -1,170 +0,0 @@
-<script lang="ts">
-	import { Plus, MessageSquare, Zap, FolderOpen } from '@lucide/svelte';
-	import { Button } from '$lib/components/ui/button';
-	import * as Sheet from '$lib/components/ui/sheet';
-	import { FILE_TYPE_ICONS } from '$lib/constants';
-	import { McpLogo } from '$lib/components/app';
-
-	interface Props {
-		class?: string;
-		disabled?: boolean;
-		hasAudioModality?: boolean;
-		hasVisionModality?: boolean;
-		hasMcpPromptsSupport?: boolean;
-		hasMcpResourcesSupport?: boolean;
-		onFileUpload?: () => void;
-		onSystemPromptClick?: () => void;
-		onMcpPromptClick?: () => void;
-		onMcpSettingsClick?: () => void;
-		onMcpResourcesClick?: () => void;
-	}
-
-	let {
-		class: className = '',
-		disabled = false,
-		hasAudioModality = false,
-		hasVisionModality = false,
-		hasMcpPromptsSupport = false,
-		hasMcpResourcesSupport = false,
-		onFileUpload,
-		onSystemPromptClick,
-		onMcpPromptClick,
-		onMcpSettingsClick,
-		onMcpResourcesClick
-	}: Props = $props();
-
-	let sheetOpen = $state(false);
-
-	function handleMcpPromptClick() {
-		sheetOpen = false;
-		onMcpPromptClick?.();
-	}
-
-	function handleMcpSettingsClick() {
-		onMcpSettingsClick?.();
-	}
-
-	function handleMcpResourcesClick() {
-		sheetOpen = false;
-		onMcpResourcesClick?.();
-	}
-
-	function handleSheetFileUpload() {
-		sheetOpen = false;
-		onFileUpload?.();
-	}
-
-	function handleSheetSystemPromptClick() {
-		sheetOpen = false;
-		onSystemPromptClick?.();
-	}
-
-	const fileUploadTooltipText = 'Add files, system prompt or MCP Servers';
-
-	const sheetItemClass =
-		'flex w-full items-center gap-3 rounded-md px-3 py-2.5 text-left text-sm transition-colors hover:bg-accent active:bg-accent disabled:cursor-not-allowed disabled:opacity-50';
-</script>
-
-<div class="flex items-center gap-1 {className}">
-	<Sheet.Root bind:open={sheetOpen}>
-		<Button
-			class="file-upload-button h-8 w-8 rounded-full p-0"
-			{disabled}
-			variant="secondary"
-			type="button"
-			onclick={() => (sheetOpen = true)}
-		>
-			<span class="sr-only">{fileUploadTooltipText}</span>
-
-			<Plus class="h-4 w-4" />
-		</Button>
-
-		<Sheet.Content side="bottom" class="max-h-[85vh] gap-0">
-			<Sheet.Header>
-				<Sheet.Title>Add to chat</Sheet.Title>
-
-				<Sheet.Description class="sr-only">
-					Add files, system prompt or configure MCP servers
-				</Sheet.Description>
-			</Sheet.Header>
-
-			<div class="flex flex-col gap-1 overflow-y-auto px-1.5 pb-2">
-				<!-- Images -->
-				<button
-					type="button"
-					class={sheetItemClass}
-					disabled={!hasVisionModality}
-					onclick={handleSheetFileUpload}
-				>
-					<FILE_TYPE_ICONS.image class="h-4 w-4 shrink-0" />
-
-					<span>Images</span>
-
-					{#if !hasVisionModality}
-						<span class="ml-auto text-xs text-muted-foreground">Requires vision model</span>
-					{/if}
-				</button>
-
-				<!-- Audio -->
-				<button
-					type="button"
-					class={sheetItemClass}
-					disabled={!hasAudioModality}
-					onclick={handleSheetFileUpload}
-				>
-					<FILE_TYPE_ICONS.audio class="h-4 w-4 shrink-0" />
-
-					<span>Audio Files</span>
-
-					{#if !hasAudioModality}
-						<span class="ml-auto text-xs text-muted-foreground">Requires audio model</span>
-					{/if}
-				</button>
-
-				<button type="button" class={sheetItemClass} onclick={handleSheetFileUpload}>
-					<FILE_TYPE_ICONS.text class="h-4 w-4 shrink-0" />
-
-					<span>Text Files</span>
-				</button>
-
-				<button type="button" class={sheetItemClass} onclick={handleSheetFileUpload}>
-					<FILE_TYPE_ICONS.pdf class="h-4 w-4 shrink-0" />
-
-					<span>PDF Files</span>
-
-					{#if !hasVisionModality}
-						<span class="ml-auto text-xs text-muted-foreground">Text-only</span>
-					{/if}
-				</button>
-
-				<button type="button" class={sheetItemClass} onclick={handleSheetSystemPromptClick}>
-					<MessageSquare class="h-4 w-4 shrink-0" />
-
-					<span>System Message</span>
-				</button>
-
-				<button type="button" class={sheetItemClass} onclick={handleMcpSettingsClick}>
-					<McpLogo class="h-4 w-4 shrink-0" />
-
-					<span>MCP Servers</span>
-				</button>
-
-				{#if hasMcpPromptsSupport}
-					<button type="button" class={sheetItemClass} onclick={handleMcpPromptClick}>
-						<Zap class="h-4 w-4 shrink-0" />
-
-						<span>MCP Prompt</span>
-					</button>
-				{/if}
-
-				{#if hasMcpResourcesSupport}
-					<button type="button" class={sheetItemClass} onclick={handleMcpResourcesClick}>
-						<FolderOpen class="h-4 w-4 shrink-0" />
-
-						<span>MCP Resources</span>
-					</button>
-				{/if}
-			</div>
-		</Sheet.Content>
-	</Sheet.Root>
-</div>

tools/server/webui/src/lib/components/app/chat/ChatForm/ChatFormActions/ChatFormActionMcpServersSubmenu.svelte

@@ -1,62 +1,39 @@
 <script lang="ts">
-	import { Settings } from '@lucide/svelte';
-	import * as DropdownMenu from '$lib/components/ui/dropdown-menu';
+	import { Settings, Plus } from '@lucide/svelte';
 	import { Switch } from '$lib/components/ui/switch';
-	import { DropdownMenuSearchable, McpActiveServersAvatars } from '$lib/components/app';
+	import * as DropdownMenu from '$lib/components/ui/dropdown-menu';
+	import { McpLogo, DropdownMenuSearchable } from '$lib/components/app';
 	import { conversationsStore } from '$lib/stores/conversations.svelte';
 	import { mcpStore } from '$lib/stores/mcp.svelte';
 	import { HealthCheckStatus } from '$lib/enums';
 	import type { MCPServerSettingsEntry } from '$lib/types';
+	import { goto } from '$app/navigation';
 
 	interface Props {
-		class?: string;
-		disabled?: boolean;
-		onSettingsClick?: () => void;
+		onMcpSettingsClick?: () => void;
 	}
 
-	let { class: className = '', disabled = false, onSettingsClick }: Props = $props();
+	let { onMcpSettingsClick }: Props = $props();
 
-	let searchQuery = $state('');
-	let mcpServers = $derived(mcpStore.getServersSorted().filter((s) => s.enabled));
+	let mcpSearchQuery = $state('');
+	let allMcpServers = $derived(mcpStore.getServersSorted());
+	let mcpServers = $derived(allMcpServers.filter((s) => s.enabled));
 	let hasMcpServers = $derived(mcpServers.length > 0);
-	let enabledMcpServersForChat = $derived(
-		mcpServers.filter((s) => conversationsStore.isMcpServerEnabledForChat(s.id) && s.url.trim())
-	);
-	let healthyEnabledMcpServers = $derived(
-		enabledMcpServersForChat.filter((s) => {
-			const healthState = mcpStore.getHealthCheckState(s.id);
-			return healthState.status !== HealthCheckStatus.ERROR;
-		})
-	);
-	let hasEnabledMcpServers = $derived(enabledMcpServersForChat.length > 0);
-	let mcpFavicons = $derived(
-		healthyEnabledMcpServers
-			.slice(0, 3)
-			.map((s) => ({ id: s.id, url: mcpStore.getServerFavicon(s.id) }))
-			.filter((f) => f.url !== null)
-	);
+	// let hasAnyMcpServers = $derived(allMcpServers.length > 0);
 	let filteredMcpServers = $derived.by(() => {
-		const query = searchQuery.toLowerCase().trim();
-		if (query) {
-			return mcpServers.filter((s) => {
-				const name = getServerLabel(s).toLowerCase();
-				const url = s.url.toLowerCase();
-				return name.includes(query) || url.includes(query);
-			});
-		}
-		return mcpServers;
+		const query = mcpSearchQuery.toLowerCase().trim();
+		if (!query) return mcpServers;
+		return mcpServers.filter((s) => {
+			const name = getServerLabel(s).toLowerCase();
+			const url = s.url.toLowerCase();
+			return name.includes(query) || url.includes(query);
+		});
 	});
 
 	function getServerLabel(server: MCPServerSettingsEntry): string {
 		return mcpStore.getServerLabel(server);
 	}
 
-	function handleDropdownOpen(open: boolean) {
-		if (open) {
-			mcpStore.runHealthChecksForServers(mcpServers);
-		}
-	}
-
 	function isServerEnabledForChat(serverId: string): boolean {
 		return conversationsStore.isMcpServerEnabledForChat(serverId);
 	}
@@ -64,38 +41,33 @@
 	async function toggleServerForChat(serverId: string) {
 		await conversationsStore.toggleMcpServerForChat(serverId);
 	}
+
+	function handleMcpSubMenuOpen(open: boolean) {
+		if (open) {
+			mcpSearchQuery = '';
+			mcpStore.runHealthChecksForServers(allMcpServers);
+		}
+	}
+
+	function handleMcpSettingsClick() {
+		onMcpSettingsClick?.();
+
+		goto(`${hasMcpServers ? '' : '?add'}#/settings/mcp`);
+	}
 </script>
 
-{#if hasMcpServers && hasEnabledMcpServers && mcpFavicons.length > 0}
-	<DropdownMenu.Root
-		onOpenChange={(open) => {
-			if (!open) {
-				searchQuery = '';
-			}
-			handleDropdownOpen(open);
-		}}
-	>
-		<DropdownMenu.Trigger
-			{disabled}
-			onclick={(e) => {
-				e.preventDefault();
-				e.stopPropagation();
-			}}
-		>
-			<button
-				type="button"
-				class="inline-flex cursor-pointer items-center rounded-sm py-1 disabled:cursor-not-allowed disabled:opacity-60"
-				{disabled}
-				aria-label="MCP Servers"
-			>
-				<McpActiveServersAvatars class={className} />
-			</button>
-		</DropdownMenu.Trigger>
+<DropdownMenu.Sub onOpenChange={handleMcpSubMenuOpen}>
+	<DropdownMenu.SubTrigger class="flex cursor-pointer items-center gap-2">
+		<McpLogo class="h-4 w-4" />
 
-		<DropdownMenu.Content align="start" class="w-72 pt-0">
+		<span>MCP Servers</span>
+	</DropdownMenu.SubTrigger>
+
+	<DropdownMenu.SubContent class="w-72 pt-0">
+		{#if hasMcpServers}
 			<DropdownMenuSearchable
-				bind:searchValue={searchQuery}
 				placeholder="Search servers..."
+				bind:searchValue={mcpSearchQuery}
 				emptyMessage="No servers found"
 				isEmpty={filteredMcpServers.length === 0}
 			>
@@ -107,7 +79,7 @@
 
 						<button
 							type="button"
-							class="flex w-full items-center justify-between gap-2 px-2 py-2 text-left transition-colors hover:bg-accent disabled:cursor-not-allowed disabled:opacity-50"
+							class="flex w-full items-center justify-between gap-2 rounded-sm px-2 py-2 text-left transition-colors hover:bg-accent disabled:cursor-not-allowed disabled:opacity-50"
 							onclick={() => !hasError && toggleServerForChat(server.id)}
 							disabled={hasError}
 						>
@@ -147,7 +119,7 @@
 				{#snippet footer()}
 					<DropdownMenu.Item
 						class="flex cursor-pointer items-center gap-2"
-						onclick={onSettingsClick}
+						onclick={handleMcpSettingsClick}
 					>
 						<Settings class="h-4 w-4" />
 
@@ -155,6 +127,21 @@
 					</DropdownMenu.Item>
 				{/snippet}
 			</DropdownMenuSearchable>
-		</DropdownMenu.Content>
-	</DropdownMenu.Root>
-{/if}
+		{:else}
+			<div class="px-2 py-3 text-center text-sm text-muted-foreground">
+				No MCP servers configured
+			</div>
+
+			<DropdownMenu.Separator />
+
+			<DropdownMenu.Item
+				class="flex cursor-pointer items-center gap-2"
+				onclick={handleMcpSettingsClick}
+			>
+				<Plus class="h-4 w-4" />
+
+				<span>Add MCP Servers</span>
+			</DropdownMenu.Item>
+		{/if}
+	</DropdownMenu.SubContent>
+</DropdownMenu.Sub>

tools/server/webui/src/lib/components/app/chat/ChatForm/ChatFormActions/ChatFormActionSubmit.svelte

@@ -7,20 +7,13 @@
 	interface Props {
 		canSend?: boolean;
 		disabled?: boolean;
-		isLoading?: boolean;
 		showErrorState?: boolean;
 		tooltipLabel?: string;
 	}
 
-	let {
-		canSend = false,
-		disabled = false,
-		isLoading = false,
-		showErrorState = false,
-		tooltipLabel
-	}: Props = $props();
+	let { canSend = false, disabled = false, showErrorState = false, tooltipLabel }: Props = $props();
 
-	let isDisabled = $derived(!canSend || disabled || isLoading);
+	let isDisabled = $derived(!canSend || disabled);
 </script>
 
 {#snippet submitButton(props = {})}

tools/server/webui/src/lib/components/app/chat/ChatForm/ChatFormActions/ChatFormActionToolsSubmenu.svelte

@@ -0,0 +1,146 @@
+<script lang="ts">
+	import { PencilRuler, ChevronDown, ChevronRight, Loader2, Info } from '@lucide/svelte';
+	import { Checkbox } from '$lib/components/ui/checkbox';
+	import * as Collapsible from '$lib/components/ui/collapsible';
+	import * as DropdownMenu from '$lib/components/ui/dropdown-menu';
+	import * as Tooltip from '$lib/components/ui/tooltip';
+	import { toolsStore } from '$lib/stores/tools.svelte';
+	import { mcpStore } from '$lib/stores/mcp.svelte';
+	import { useToolsPanel } from '$lib/hooks/use-tools-panel.svelte';
+
+	const toolsPanel = useToolsPanel();
+	const hasMcpServersAvailable = $derived(mcpStore.getServersSorted().length > 0);
+</script>
+
+<DropdownMenu.Sub onOpenChange={(open) => open && toolsPanel.handleOpen()}>
+	<DropdownMenu.SubTrigger class="flex cursor-pointer items-center gap-2">
+		<PencilRuler class="h-4 w-4" />
+
+		<span>Tools</span>
+	</DropdownMenu.SubTrigger>
+
+	<DropdownMenu.SubContent class="w-72 p-0">
+		{#if toolsPanel.totalToolCount === 0}
+			{#if toolsStore.loading}
+				<div class="px-3 py-4 text-center text-sm text-muted-foreground">
+					<Loader2 class="mx-auto mb-1 h-4 w-4 animate-spin" />
+					Loading tools...
+				</div>
+			{:else if toolsStore.isToolsEndpointUnreachable}
+				<div class="grid gap-2.5 px-3 py-4 text-sm text-muted-foreground">
+					<span class="flex gap-2">
+						<Info class="mt-0.5 h-4 w-4 shrink-0" />
+
+						<span
+							>Run llama-server with <code>--tools</code> flag to enable
+							<strong>Built-in Tools</strong>.</span
+						>
+					</span>
+
+					<span class="flex gap-2">
+						<Info class="mt-0.5 h-4 w-4 shrink-0" />
+
+						<span
+							>{hasMcpServersAvailable ? 'Enable' : 'Add'} MCP Server(s) to access
+							<strong>MCP Tools</strong>.</span
+						>
+					</span>
+				</div>
+			{:else if toolsStore.error}
+				<div class="px-3 py-4 text-center text-sm text-muted-foreground">Failed to load tools</div>
+			{:else if toolsPanel.noToolsInfoMessage}
+				<div class="flex gap-2 px-3 py-4 text-sm text-muted-foreground">
+					<Info class="mt-0.5 h-4 w-4 shrink-0" />
+
+					<span>{toolsPanel.noToolsInfoMessage}</span>
+				</div>
+			{:else}
+				<div class="px-3 py-4 text-center text-sm text-muted-foreground">No tools available</div>
+			{/if}
+		{:else}
+			<div class="max-h-80 overflow-y-auto p-2 pr-1">
+				{#each toolsPanel.activeGroups as group (group.label)}
+					{@const isExpanded = toolsPanel.expandedGroups.has(group.label)}
+					{@const { checked, indeterminate } = toolsPanel.getGroupCheckedState(group)}
+					{@const favicon = toolsPanel.getFavicon(group)}
+
+					<Collapsible.Root
+						open={isExpanded}
+						onOpenChange={() => toolsPanel.toggleGroupExpanded(group.label)}
+					>
+						<div class="flex items-center gap-1">
+							<Collapsible.Trigger
+								class="flex min-w-0 flex-1 items-center gap-2 rounded px-2 py-1.5 text-sm hover:bg-muted/50"
+							>
+								{#if isExpanded}
+									<ChevronDown class="h-3.5 w-3.5 shrink-0" />
+								{:else}
+									<ChevronRight class="h-3.5 w-3.5 shrink-0" />
+								{/if}
+
+								<span class="inline-flex min-w-0 items-center gap-1.5 font-medium">
+									{#if favicon}
+										<img
+											src={favicon}
+											alt=""
+											class="h-4 w-4 shrink-0 rounded-sm"
+											onerror={(e) => {
+												(e.currentTarget as HTMLImageElement).style.display = 'none';
+											}}
+										/>
+									{/if}
+
+									<span class="truncate">{group.label}</span>
+								</span>
+
+								<span class="ml-auto shrink-0 text-xs text-muted-foreground">
+									{toolsPanel.getEnabledToolCount(group)}/{group.tools.length}
+								</span>
+							</Collapsible.Trigger>
+
+							<Tooltip.Root>
+								<Tooltip.Trigger>
+									<Checkbox
+										{checked}
+										{indeterminate}
+										onCheckedChange={() => toolsStore.toggleGroup(group)}
+										class="mr-2 h-4 w-4 shrink-0"
+									/>
+								</Tooltip.Trigger>
+
+								<Tooltip.Content side="right">
+									<p>
+										{checked ? 'Disable' : 'Enable'}
+										{group.tools.length} tool{group.tools.length !== 1 ? 's' : ''}
+									</p>
+								</Tooltip.Content>
+							</Tooltip.Root>
+						</div>
+
+						<Collapsible.Content>
+							<div class="ml-4 flex flex-col gap-0.5 border-l border-border/50 pl-2">
+								{#each group.tools as tool (tool.function.name)}
+									<button
+										type="button"
+										class="flex w-full items-center gap-2 rounded px-2 py-1.5 text-left text-sm transition-colors hover:bg-muted/50"
+										onclick={() => toolsStore.toggleTool(tool.function.name)}
+									>
+										<Checkbox
+											checked={toolsStore.isToolEnabled(tool.function.name)}
+											onCheckedChange={() => toolsStore.toggleTool(tool.function.name)}
+											class="h-4 w-4 shrink-0"
+										/>
+
+										<span class="min-w-0 flex-1 truncate font-mono text-[12px]">
+											{tool.function.name}
+										</span>
+									</button>
+								{/each}
+							</div>
+						</Collapsible.Content>
+					</Collapsible.Root>
+				{/each}
+			</div>
+		{/if}
+	</DropdownMenu.SubContent>
+</DropdownMenu.Sub>

tools/server/webui/src/lib/components/app/chat/ChatForm/ChatFormActions/ChatFormActions.svelte

@@ -6,21 +6,19 @@
 		ChatFormActionAttachmentsSheet,
 		ChatFormActionRecord,
 		ChatFormActionSubmit,
-		McpServersSelector,
-		ModelsSelector,
+		ModelsSelectorDropdown,
 		ModelsSelectorSheet
 	} from '$lib/components/app';
-	import { SETTINGS_SECTION_TITLES } from '$lib/constants';
-	import { mcpStore } from '$lib/stores/mcp.svelte';
-	import { getChatSettingsDialogContext } from '$lib/contexts';
 	import { FileTypeCategory } from '$lib/enums';
-	import { getFileTypeCategory } from '$lib/utils';
-	import { config } from '$lib/stores/settings.svelte';
+	import { IsMobile } from '$lib/hooks/is-mobile.svelte';
+	import { chatStore } from '$lib/stores/chat.svelte';
+	import { mcpStore } from '$lib/stores/mcp.svelte';
 	import { modelsStore, modelOptions, selectedModelId } from '$lib/stores/models.svelte';
 	import { isRouterMode, serverError } from '$lib/stores/server.svelte';
-	import { chatStore } from '$lib/stores/chat.svelte';
+	import { config } from '$lib/stores/settings.svelte';
 	import { activeMessages, conversationsStore } from '$lib/stores/conversations.svelte';
-	import { IsMobile } from '$lib/hooks/is-mobile.svelte';
+	import { getFileTypeCategory } from '$lib/utils';
+	import { goto } from '$app/navigation';
 
 	interface Props {
 		canSend?: boolean;
@@ -165,7 +163,8 @@
 		return '';
 	});
 
-	let selectorModelRef: ModelsSelector | ModelsSelectorSheet | undefined = $state(undefined);
+	let selectorModelRef: ModelsSelectorDropdown | ModelsSelectorSheet | undefined =
+		$state(undefined);
 
 	let isMobile = new IsMobile();
 
@@ -173,8 +172,6 @@
 		selectorModelRef?.open();
 	}
 
-	const chatSettingsDialog = getChatSettingsDialogContext();
-
 	let hasMcpPromptsSupport = $derived.by(() => {
 		const perChatOverrides = conversationsStore.getAllMcpServerOverrides();
 
@@ -200,8 +197,8 @@
 				{onFileUpload}
 				{onSystemPromptClick}
 				{onMcpPromptClick}
+				onMcpSettingsClick={() => goto('#/settings/mcp')}
 				{onMcpResourcesClick}
-				onMcpSettingsClick={() => chatSettingsDialog.open(SETTINGS_SECTION_TITLES.MCP)}
 			/>
 		{:else}
 			<ChatFormActionAttachmentsDropdown
@@ -214,17 +211,12 @@
 				{onSystemPromptClick}
 				{onMcpPromptClick}
 				{onMcpResourcesClick}
-				onMcpSettingsClick={() => chatSettingsDialog.open(SETTINGS_SECTION_TITLES.MCP)}
+				onMcpSettingsClick={() => goto('#/settings/mcp')}
 			/>
 		{/if}
-
-		<McpServersSelector
-			{disabled}
-			onSettingsClick={() => chatSettingsDialog.open(SETTINGS_SECTION_TITLES.MCP)}
-		/>
 	</div>
 
-	<div class="ml-auto flex items-center gap-1.5">
+	<div class="ml-auto flex items-center gap-2">
 		{#if isMobile.current}
 			<ModelsSelectorSheet
 				disabled={disabled || isOffline}
@@ -234,7 +226,7 @@
 				useGlobalSelection
 			/>
 		{:else}
-			<ModelsSelector
+			<ModelsSelectorDropdown
 				disabled={disabled || isOffline}
 				bind:this={selectorModelRef}
 				currentModel={conversationModel}
@@ -244,7 +236,7 @@
 		{/if}
 	</div>
 
-	{#if isLoading}
+	{#if isLoading && !hasText}
 		<Button
 			type="button"
 			variant="secondary"
@@ -263,7 +255,6 @@
 		<ChatFormActionSubmit
 			canSend={canSend && hasModelSelected && isSelectedModelInCache}
 			{disabled}
-			{isLoading}
 			tooltipLabel={submitTooltip}
 			showErrorState={hasModelSelected && !isSelectedModelInCache}
 		/>

tools/server/webui/src/lib/components/app/chat/ChatForm/ChatFormActions/ChatFormActionsAttachments/ChatFormActionAttachmentsDropdown.svelte

@@ -0,0 +1,182 @@
+<script lang="ts">
+	import { Plus } from '@lucide/svelte';
+	import { Button } from '$lib/components/ui/button';
+	import * as DropdownMenu from '$lib/components/ui/dropdown-menu';
+	import * as Tooltip from '$lib/components/ui/tooltip';
+	import {
+		ATTACHMENT_FILE_ITEMS,
+		ATTACHMENT_EXTRA_ITEMS,
+		ATTACHMENT_MCP_ITEMS,
+		ATTACHMENT_TOOLTIP_TEXT,
+		TOOLTIP_DELAY_DURATION
+	} from '$lib/constants';
+	import { AttachmentMenuItemId } from '$lib/enums';
+	import { ChatFormActionToolsSubmenu, ChatFormActionMcpServersSubmenu } from '$lib/components/app';
+
+	import { useAttachmentMenu } from '$lib/hooks/use-attachment-menu.svelte';
+
+	interface Props {
+		class?: string;
+		disabled?: boolean;
+		hasAudioModality?: boolean;
+		hasVisionModality?: boolean;
+		hasMcpPromptsSupport?: boolean;
+		hasMcpResourcesSupport?: boolean;
+		onFileUpload?: () => void;
+		onSystemPromptClick?: () => void;
+		onMcpPromptClick?: () => void;
+		onMcpSettingsClick?: () => void;
+		onMcpResourcesClick?: () => void;
+	}
+
+	let {
+		class: className = '',
+		disabled = false,
+		hasAudioModality = false,
+		hasVisionModality = false,
+		hasMcpPromptsSupport = false,
+		hasMcpResourcesSupport = false,
+		onFileUpload,
+		onSystemPromptClick,
+		onMcpPromptClick,
+		onMcpSettingsClick,
+		onMcpResourcesClick
+	}: Props = $props();
+
+	let dropdownOpen = $state(false);
+
+	function handleMcpSettingsClick() {
+		dropdownOpen = false;
+		onMcpSettingsClick?.();
+	}
+
+	const attachmentMenu = useAttachmentMenu(
+		() => ({ hasVisionModality, hasAudioModality, hasMcpPromptsSupport, hasMcpResourcesSupport }),
+		() => ({ onFileUpload, onSystemPromptClick, onMcpPromptClick, onMcpResourcesClick }),
+		() => {
+			dropdownOpen = false;
+		}
+	);
+</script>
+
+<div class="flex items-center gap-1 {className}">
+	<DropdownMenu.Root bind:open={dropdownOpen}>
+		<DropdownMenu.Trigger name="Attach files" {disabled}>
+			<Tooltip.Root>
+				<Tooltip.Trigger class="w-full">
+					<Button
+						class="file-upload-button h-8 w-8 rounded-full p-0"
+						{disabled}
+						variant="secondary"
+						type="button"
+					>
+						<span class="sr-only">{ATTACHMENT_TOOLTIP_TEXT}</span>
+
+						<Plus class="h-4 w-4" />
+					</Button>
+				</Tooltip.Trigger>
+
+				<Tooltip.Content>
+					<p>{ATTACHMENT_TOOLTIP_TEXT}</p>
+				</Tooltip.Content>
+			</Tooltip.Root>
+		</DropdownMenu.Trigger>
+
+		<DropdownMenu.Content align="start" class="w-48">
+			{#each ATTACHMENT_FILE_ITEMS as item (item.id)}
+				{@const enabled = attachmentMenu.isItemEnabled(item.enabledWhen)}
+				{#if enabled}
+					<DropdownMenu.Item
+						class="{item.class ?? ''} flex cursor-pointer items-center gap-2"
+						onclick={() => attachmentMenu.callbacks[item.action]()}
+					>
+						<item.icon class="h-4 w-4" />
+
+						<span>{item.label}</span>
+					</DropdownMenu.Item>
+				{:else if item.disabledTooltip}
+					<Tooltip.Root delayDuration={TOOLTIP_DELAY_DURATION}>
+						<Tooltip.Trigger class="w-full">
+							<DropdownMenu.Item
+								class="{item.class ?? ''} flex cursor-pointer items-center gap-2"
+								disabled
+							>
+								<item.icon class="h-4 w-4" />
+
+								<span>{item.label}</span>
+							</DropdownMenu.Item>
+						</Tooltip.Trigger>
+
+						<Tooltip.Content side="right">
+							<p>{item.disabledTooltip}</p>
+						</Tooltip.Content>
+					</Tooltip.Root>
+				{/if}
+			{/each}
+
+			{#if !attachmentMenu.isItemEnabled('hasVisionModality')}
+				<Tooltip.Root delayDuration={TOOLTIP_DELAY_DURATION}>
+					<Tooltip.Trigger class="w-full">
+						<DropdownMenu.Item
+							class="flex cursor-pointer items-center gap-2"
+							onclick={attachmentMenu.callbacks.onFileUpload}
+						>
+							{@const pdfItem = ATTACHMENT_FILE_ITEMS.find(
+								(i) => i.id === AttachmentMenuItemId.PDF
+							)}
+							{#if pdfItem}
+								<pdfItem.icon class="h-4 w-4" />
+
+								<span>{pdfItem.label}</span>
+							{/if}
+						</DropdownMenu.Item>
+					</Tooltip.Trigger>
+
+					<Tooltip.Content side="right">
+						<p>PDFs will be converted to text. Image-based PDFs may not work properly.</p>
+					</Tooltip.Content>
+				</Tooltip.Root>
+			{/if}
+
+			<DropdownMenu.Separator />
+
+			{#each ATTACHMENT_EXTRA_ITEMS as item (item.id)}
+				{#if item.id === AttachmentMenuItemId.SYSTEM_MESSAGE}
+					<Tooltip.Root delayDuration={TOOLTIP_DELAY_DURATION}>
+						<Tooltip.Trigger class="w-full">
+							<DropdownMenu.Item
+								class="flex cursor-pointer items-center gap-2"
+								onclick={() => attachmentMenu.callbacks[item.action]()}
+							>
+								<item.icon class="h-4 w-4" />
+
+								<span>{item.label}</span>
+							</DropdownMenu.Item>
+						</Tooltip.Trigger>
+
+						<Tooltip.Content side="right">
+							<p>{attachmentMenu.getSystemMessageTooltip()}</p>
+						</Tooltip.Content>
+					</Tooltip.Root>
+				{/if}
+			{/each}
+
+			<ChatFormActionToolsSubmenu />
+
+			<ChatFormActionMcpServersSubmenu onMcpSettingsClick={handleMcpSettingsClick} />
+
+			{#each ATTACHMENT_MCP_ITEMS as item (item.id)}
+				{#if attachmentMenu.isItemVisible(item.visibleWhen)}
+					<DropdownMenu.Item
+						class="flex cursor-pointer items-center gap-2"
+						onclick={() => attachmentMenu.callbacks[item.action]()}
+					>
+						<item.icon class="h-4 w-4" />
+
+						<span>{item.label}</span>
+					</DropdownMenu.Item>
+				{/if}
+			{/each}
+		</DropdownMenu.Content>
+	</DropdownMenu.Root>
+</div>

tools/server/webui/src/lib/components/app/chat/ChatForm/ChatFormActions/ChatFormActionsAttachments/ChatFormActionAttachmentsSheet.svelte

@@ -0,0 +1,184 @@
+<script lang="ts">
+	import { Plus } from '@lucide/svelte';
+	import { Button } from '$lib/components/ui/button';
+	import * as Tooltip from '$lib/components/ui/tooltip';
+	import * as Sheet from '$lib/components/ui/sheet';
+	import { TOOLTIP_DELAY_DURATION } from '$lib/constants';
+	import {
+		ATTACHMENT_FILE_ITEMS,
+		ATTACHMENT_EXTRA_ITEMS,
+		ATTACHMENT_MCP_ITEMS,
+		ATTACHMENT_TOOLTIP_TEXT
+	} from '$lib/constants/attachment-menu';
+	import { ChatFormActionToolsSubmenu, ChatFormActionMcpServersSubmenu } from '$lib/components/app';
+	import { useAttachmentMenu } from '$lib/hooks/use-attachment-menu.svelte';
+	import { AttachmentMenuItemId } from '$lib/enums';
+
+	interface Props {
+		class?: string;
+		disabled?: boolean;
+		hasAudioModality?: boolean;
+		hasVisionModality?: boolean;
+		hasMcpPromptsSupport?: boolean;
+		hasMcpResourcesSupport?: boolean;
+		onFileUpload?: () => void;
+		onSystemPromptClick?: () => void;
+		onMcpPromptClick?: () => void;
+		onMcpSettingsClick?: () => void;
+		onMcpResourcesClick?: () => void;
+	}
+
+	let {
+		class: className = '',
+		disabled = false,
+		hasAudioModality = false,
+		hasVisionModality = false,
+		hasMcpPromptsSupport = false,
+		hasMcpResourcesSupport = false,
+		onFileUpload,
+		onSystemPromptClick,
+		onMcpPromptClick,
+		onMcpSettingsClick,
+		onMcpResourcesClick
+	}: Props = $props();
+
+	let sheetOpen = $state(false);
+
+	const attachmentMenu = useAttachmentMenu(
+		() => ({ hasVisionModality, hasAudioModality, hasMcpPromptsSupport, hasMcpResourcesSupport }),
+		() => ({ onFileUpload, onSystemPromptClick, onMcpPromptClick, onMcpResourcesClick }),
+		() => {
+			sheetOpen = false;
+		}
+	);
+
+	function handleMcpSettingsClick() {
+		sheetOpen = false;
+		onMcpSettingsClick?.();
+	}
+
+	const sheetItemClass =
+		'flex w-full items-center gap-3 rounded-md px-3 py-2.5 text-left text-sm transition-colors hover:bg-accent active:bg-accent disabled:cursor-not-allowed disabled:opacity-50';
+</script>
+
+<div class="flex items-center gap-1 {className}">
+	<Sheet.Root bind:open={sheetOpen}>
+		<Button
+			class="file-upload-button h-8 w-8 rounded-full p-0"
+			{disabled}
+			variant="secondary"
+			type="button"
+			onclick={() => (sheetOpen = true)}
+		>
+			<span class="sr-only">{ATTACHMENT_TOOLTIP_TEXT}</span>
+
+			<Plus class="h-4 w-4" />
+		</Button>
+
+		<Sheet.Content side="bottom" class="max-h-[85vh] gap-0 overflow-y-auto">
+			<Sheet.Header>
+				<Sheet.Title>Add to chat</Sheet.Title>
+
+				<Sheet.Description class="sr-only">
+					Add files, system prompt or configure MCP servers
+				</Sheet.Description>
+			</Sheet.Header>
+
+			<div class="flex flex-col gap-1 px-1.5 pb-2">
+				{#each ATTACHMENT_FILE_ITEMS as item (item.id)}
+					{@const enabled = attachmentMenu.isItemEnabled(item.enabledWhen)}
+					{#if enabled}
+						<button
+							type="button"
+							class={sheetItemClass}
+							onclick={() => attachmentMenu.callbacks[item.action]()}
+						>
+							<item.icon class="h-4 w-4 shrink-0" />
+
+							<span>{item.label}</span>
+						</button>
+					{:else if item.disabledTooltip}
+						<Tooltip.Root delayDuration={TOOLTIP_DELAY_DURATION}>
+							<Tooltip.Trigger>
+								<button type="button" class={sheetItemClass} disabled>
+									<item.icon class="h-4 w-4 shrink-0" />
+
+									<span>{item.label}</span>
+								</button>
+							</Tooltip.Trigger>
+
+							<Tooltip.Content side="right">
+								<p>{item.disabledTooltip}</p>
+							</Tooltip.Content>
+						</Tooltip.Root>
+					{/if}
+				{/each}
+
+				{#if !attachmentMenu.isItemEnabled('hasVisionModality')}
+					{@const pdfItem = ATTACHMENT_FILE_ITEMS.find((i) => i.id === AttachmentMenuItemId.PDF)}
+					{#if pdfItem}
+						<Tooltip.Root delayDuration={TOOLTIP_DELAY_DURATION}>
+							<Tooltip.Trigger>
+								<button
+									type="button"
+									class={sheetItemClass}
+									onclick={() => attachmentMenu.callbacks[pdfItem.action]()}
+								>
+									<pdfItem.icon class="h-4 w-4 shrink-0" />
+
+									<span>{pdfItem.label}</span>
+								</button>
+							</Tooltip.Trigger>
+
+							<Tooltip.Content side="right">
+								<p>PDFs will be converted to text. Image-based PDFs may not work properly.</p>
+							</Tooltip.Content>
+						</Tooltip.Root>
+					{/if}
+				{/if}
+
+				{#each ATTACHMENT_EXTRA_ITEMS as item (item.id)}
+					{#if item.id === AttachmentMenuItemId.SYSTEM_MESSAGE}
+						<Tooltip.Root delayDuration={TOOLTIP_DELAY_DURATION}>
+							<Tooltip.Trigger>
+								<button
+									type="button"
+									class={sheetItemClass}
+									onclick={() => attachmentMenu.callbacks[item.action]()}
+								>
+									<item.icon class="h-4 w-4 shrink-0" />
+
+									<span>{item.label}</span>
+								</button>
+							</Tooltip.Trigger>
+
+							<Tooltip.Content side="right">
+								<p>{attachmentMenu.getSystemMessageTooltip()}</p>
+							</Tooltip.Content>
+						</Tooltip.Root>
+					{/if}
+				{/each}
+
+				<div class="my-2 border-t"></div>
+
+				<ChatFormActionToolsSubmenu />
+
+				<ChatFormActionMcpServersSubmenu onMcpSettingsClick={handleMcpSettingsClick} />
+
+				{#each ATTACHMENT_MCP_ITEMS as item (item.id)}
+					{#if attachmentMenu.isItemVisible(item.visibleWhen)}
+						<button
+							type="button"
+							class={sheetItemClass}
+							onclick={() => attachmentMenu.callbacks[item.action]()}
+						>
+							<item.icon class="h-4 w-4 shrink-0" />
+
+							<span>{item.label}</span>
+						</button>
+					{/if}
+				{/each}
+			</div>
+		</Sheet.Content>
+	</Sheet.Root>
+</div>

tools/server/webui/src/lib/components/app/chat/ChatMessages/ChatMessage.svelte

@@ -1,12 +1,12 @@
 <script lang="ts">
 	import { goto } from '$app/navigation';
-	import { base } from '$app/paths';
 	import { getChatActionsContext, setMessageEditContext } from '$lib/contexts';
 	import { chatStore, pendingEditMessageId } from '$lib/stores/chat.svelte';
 	import { conversationsStore } from '$lib/stores/conversations.svelte';
 	import { DatabaseService } from '$lib/services/database.service';
 	import { SYSTEM_MESSAGE_PLACEHOLDER } from '$lib/constants';
 	import { MessageRole, AttachmentType } from '$lib/enums';
+	import { fadeInView } from '$lib/actions/fade-in-view.svelte';
 	import {
 		ChatMessageAssistant,
 		ChatMessageUser,
@@ -118,7 +118,7 @@
 			const conversationDeleted = await chatStore.removeSystemPromptPlaceholder(message.id);
 
 			if (conversationDeleted) {
-				goto(`${base}/`);
+				goto(`#/`);
 			}
 
 			return;
@@ -138,7 +138,7 @@
 			const conversationDeleted = await chatStore.removeSystemPromptPlaceholder(message.id);
 
 			if (conversationDeleted) {
-				goto(`${base}/`);
+				goto(`#/`);
 			}
 		} else {
 			chatActions.delete(message);
@@ -200,7 +200,7 @@
 				const conversationDeleted = await chatStore.removeSystemPromptPlaceholder(message.id);
 				isEditing = false;
 				if (conversationDeleted) {
-					goto(`${base}/`);
+					goto(`#/`);
 				}
 				return;
 			}
@@ -252,70 +252,72 @@
 	}
 </script>
 
-{#if message.role === MessageRole.SYSTEM}
-	<ChatMessageSystem
-		bind:textareaElement
-		class={className}
-		{deletionInfo}
-		{message}
-		onConfirmDelete={handleConfirmDelete}
-		onCopy={handleCopy}
-		onDelete={handleDelete}
-		onEdit={handleEdit}
-		onNavigateToSibling={handleNavigateToSibling}
-		onShowDeleteDialogChange={handleShowDeleteDialogChange}
-		{showDeleteDialog}
-		{siblingInfo}
-	/>
-{:else if mcpPromptExtra}
-	<ChatMessageMcpPrompt
-		class={className}
-		{deletionInfo}
-		{message}
-		mcpPrompt={mcpPromptExtra}
-		onConfirmDelete={handleConfirmDelete}
-		onCopy={handleCopy}
-		onDelete={handleDelete}
-		onEdit={handleEdit}
-		onNavigateToSibling={handleNavigateToSibling}
-		onShowDeleteDialogChange={handleShowDeleteDialogChange}
-		{showDeleteDialog}
-		{siblingInfo}
-	/>
-{:else if message.role === MessageRole.USER}
-	<ChatMessageUser
-		class={className}
-		{deletionInfo}
-		{message}
-		onConfirmDelete={handleConfirmDelete}
-		onCopy={handleCopy}
-		onDelete={handleDelete}
-		onEdit={handleEdit}
-		onForkConversation={handleForkConversation}
-		onNavigateToSibling={handleNavigateToSibling}
-		onShowDeleteDialogChange={handleShowDeleteDialogChange}
-		{showDeleteDialog}
-		{siblingInfo}
-	/>
-{:else}
-	<ChatMessageAssistant
-		bind:textareaElement
-		class={className}
-		{deletionInfo}
-		{isLastAssistantMessage}
-		{message}
-		{toolMessages}
-		messageContent={message.content}
-		onConfirmDelete={handleConfirmDelete}
-		onContinue={handleContinue}
-		onCopy={handleCopy}
-		onDelete={handleDelete}
-		onEdit={handleEdit}
-		onForkConversation={handleForkConversation}
-		onNavigateToSibling={handleNavigateToSibling}
-		onRegenerate={handleRegenerate}
-		onShowDeleteDialogChange={handleShowDeleteDialogChange}
-		{showDeleteDialog}
-		{siblingInfo}
-	/>
-{/if}
+<div use:fadeInView>
+	{#if message.role === MessageRole.SYSTEM}
+		<ChatMessageSystem
+			bind:textareaElement
+			class={className}
+			{deletionInfo}
+			{message}
+			onConfirmDelete={handleConfirmDelete}
+			onCopy={handleCopy}
+			onDelete={handleDelete}
+			onEdit={handleEdit}
+			onNavigateToSibling={handleNavigateToSibling}
+			onShowDeleteDialogChange={handleShowDeleteDialogChange}
+			{showDeleteDialog}
+			{siblingInfo}
+		/>
+	{:else if mcpPromptExtra}
+		<ChatMessageMcpPrompt
+			class={className}
+			{deletionInfo}
+			{message}
+			mcpPrompt={mcpPromptExtra}
+			onConfirmDelete={handleConfirmDelete}
+			onCopy={handleCopy}
+			onDelete={handleDelete}
+			onEdit={handleEdit}
+			onNavigateToSibling={handleNavigateToSibling}
+			onShowDeleteDialogChange={handleShowDeleteDialogChange}
+			{showDeleteDialog}
+			{siblingInfo}
+		/>
+	{:else if message.role === MessageRole.USER}
+		<ChatMessageUser
+			class={className}
+			{deletionInfo}
+			{message}
+			onConfirmDelete={handleConfirmDelete}
+			onCopy={handleCopy}
+			onDelete={handleDelete}
+			onEdit={handleEdit}
+			onForkConversation={handleForkConversation}
+			onNavigateToSibling={handleNavigateToSibling}
+			onShowDeleteDialogChange={handleShowDeleteDialogChange}
+			{showDeleteDialog}
+			{siblingInfo}
+		/>
+	{:else}
+		<ChatMessageAssistant
+			bind:textareaElement
+			class={className}
+			{deletionInfo}
+			{isLastAssistantMessage}
+			{message}
+			{toolMessages}
+			messageContent={message.content}
+			onConfirmDelete={handleConfirmDelete}
+			onContinue={handleContinue}
+			onCopy={handleCopy}
+			onDelete={handleDelete}
+			onEdit={handleEdit}
+			onForkConversation={handleForkConversation}
+			onNavigateToSibling={handleNavigateToSibling}
+			onRegenerate={handleRegenerate}
+			onShowDeleteDialogChange={handleShowDeleteDialogChange}
+			{showDeleteDialog}
+			{siblingInfo}
+		/>
+	{/if}
+</div>

tools/server/webui/src/lib/components/app/chat/ChatMessages/ChatMessageActionCard.svelte

@@ -0,0 +1,23 @@
+<script lang="ts">
+	import type { Snippet, Component } from 'svelte';
+
+	interface Props {
+		icon: Component<{ class?: string }>;
+		message: Snippet;
+		actions: Snippet;
+	}
+
+	let { icon: Icon, message, actions }: Props = $props();
+</script>
+
+<div class="my-2 rounded-lg border border-border bg-card p-3">
+	<div class="mb-3 flex items-center gap-2 text-sm">
+		<Icon class="h-4 w-4 shrink-0 text-muted-foreground" />
+		<span>
+			{@render message()}
+		</span>
+	</div>
+	<div class="flex flex-wrap items-center gap-2">
+		{@render actions()}
+	</div>
+</div>

tools/server/webui/src/lib/components/app/chat/ChatMessages/ChatMessageAgenticContent.svelte

@@ -1,38 +1,104 @@
 <script lang="ts">
+	import { Wrench, Loader2, Brain } from '@lucide/svelte';
 	import {
 		ChatMessageStatistics,
 		CollapsibleContentBlock,
 		MarkdownContent,
-		SyntaxHighlightedCode
+		SyntaxHighlightedCode,
+		ChatMessagePermissionRequest,
+		ChatMessageContinueRequest
 	} from '$lib/components/app';
-	import { config } from '$lib/stores/settings.svelte';
-	import { Wrench, Loader2, Brain } from '@lucide/svelte';
-	import { AgenticSectionType, FileTypeText } from '$lib/enums';
-	import { formatJsonPretty } from '$lib/utils';
+
+	import {
+		AgenticSectionType,
+		ChatMessageStatsView,
+		FileTypeText,
+		ToolPermissionDecision
+	} from '$lib/enums';
+	import type {
+		ChatMessageAgenticTimings,
+		ChatMessageAgenticTurnStats,
+		DatabaseMessage
+	} from '$lib/types';
 	import {
 		deriveAgenticSections,
+		formatJsonPretty,
 		parseToolResultWithImages,
 		type AgenticSection,
 		type ToolResultLine
 	} from '$lib/utils';
-	import type { DatabaseMessage } from '$lib/types/database';
-	import type { ChatMessageAgenticTimings, ChatMessageAgenticTurnStats } from '$lib/types/chat';
-	import { ChatMessageStatsView } from '$lib/enums';
+	import {
+		agenticPendingPermissionRequest,
+		agenticResolvePermission,
+		agenticPendingContinueRequest,
+		agenticResolveContinue
+	} from '$lib/stores/agentic.svelte';
+	import { config } from '$lib/stores/settings.svelte';
 
 	interface Props {
 		message: DatabaseMessage;
 		toolMessages?: DatabaseMessage[];
 		isStreaming?: boolean;
+		isLastAssistantMessage?: boolean;
 		highlightTurns?: boolean;
 	}
 
-	let { message, toolMessages = [], isStreaming = false, highlightTurns = false }: Props = $props();
+	let {
+		message,
+		toolMessages = [],
+		isStreaming = false,
+		isLastAssistantMessage = false,
+		highlightTurns = false
+	}: Props = $props();
 
 	let expandedStates: Record<number, boolean> = $state({});
 
 	const showToolCallInProgress = $derived(config().showToolCallInProgress as boolean);
 	const showThoughtInProgress = $derived(config().showThoughtInProgress as boolean);
 
+	let permissionDismissed = $state(false);
+
+	const pendingPermission = $derived(
+		isStreaming && isLastAssistantMessage ? agenticPendingPermissionRequest(message.convId) : null
+	);
+
+	// Reset dismissed when pendingPermission changes (new request or cleared)
+	let prevPendingRef: typeof pendingPermission = null;
+	$effect(() => {
+		if (pendingPermission !== prevPendingRef) {
+			prevPendingRef = pendingPermission;
+			if (pendingPermission) {
+				permissionDismissed = false;
+			}
+		}
+	});
+
+	function handlePermission(decision: ToolPermissionDecision) {
+		permissionDismissed = true;
+		agenticResolvePermission(message.convId, decision);
+	}
+
+	let continueDismissed = $state(false);
+
+	const pendingContinue = $derived(
+		isStreaming && isLastAssistantMessage ? agenticPendingContinueRequest(message.convId) : false
+	);
+
+	let prevContinueRef = false;
+	$effect(() => {
+		if (pendingContinue !== prevContinueRef) {
+			prevContinueRef = pendingContinue;
+			if (pendingContinue) {
+				continueDismissed = false;
+			}
+		}
+	});
+
+	function handleContinue(shouldContinue: boolean) {
+		continueDismissed = true;
+		agenticResolveContinue(message.convId, shouldContinue);
+	}
+
 	const sections = $derived(deriveAgenticSections(message, toolMessages, [], isStreaming));
 
 	// Parse tool results with images
@@ -201,7 +267,11 @@
 						<Loader2 class="h-3 w-3 animate-spin" />
 					{/if}
 				</div>
-				{#if section.toolResult}
+				{#if isPending}
+					<div class="rounded bg-muted/30 p-2 text-xs text-muted-foreground italic">
+						Waiting for result...
+					</div>
+				{:else if section.toolResult}
 					<div class="overflow-auto rounded-lg border border-border bg-muted p-4">
 						{#each section.parsedLines as line, i (i)}
 							<div class="font-mono text-xs leading-relaxed whitespace-pre-wrap">{line.text}</div>
@@ -215,10 +285,8 @@
 							{/if}
 						{/each}
 					</div>
-				{:else if isPending}
-					<div class="rounded bg-muted/30 p-2 text-xs text-muted-foreground italic">
-						Waiting for result...
-					</div>
+				{:else}
+					<div class="rounded bg-muted/30 p-2 text-xs text-muted-foreground italic">No output</div>
 				{/if}
 			</div>
 		</CollapsibleContentBlock>
@@ -289,6 +357,18 @@
 			{@render renderSection(section, index)}
 		{/each}
 	{/if}
+
+	{#if pendingPermission && !permissionDismissed}
+		<ChatMessagePermissionRequest
+			toolName={pendingPermission.toolName}
+			serverLabel={pendingPermission.serverLabel}
+			onDecision={handlePermission}
+		/>
+	{/if}
+
+	{#if pendingContinue && !continueDismissed}
+		<ChatMessageContinueRequest onDecision={handleContinue} />
+	{/if}
 </div>
 
 <style>

tools/server/webui/src/lib/components/app/chat/ChatMessages/ChatMessageAssistant.svelte

@@ -4,7 +4,7 @@
 		ChatMessageActions,
 		ChatMessageStatistics,
 		ModelBadge,
-		ModelsSelector
+		ModelsSelectorDropdown
 	} from '$lib/components/app';
 	import { getMessageEditContext } from '$lib/contexts';
 	import { useProcessingState } from '$lib/hooks/use-processing-state.svelte';
@@ -308,6 +308,7 @@
 				{message}
 				{toolMessages}
 				isStreaming={isChatStreaming()}
+				{isLastAssistantMessage}
 				highlightTurns={highlightAgenticTurns}
 			/>
 		{/if}
@@ -336,10 +337,10 @@
 				class="inline-flex flex-wrap items-start gap-2 text-xs text-muted-foreground"
 			>
 				{#if isRouter}
-					<ModelsSelector
+					<ModelsSelectorDropdown
 						currentModel={displayedModel}
 						disabled={isLoading()}
-						onModelChange={async (modelId, modelName) => {
+						onModelChange={async (modelId: string, modelName: string) => {
 							const status = modelsStore.getModelStatus(modelId);
 
 							if (status !== ServerModelStatus.LOADED) {

tools/server/webui/src/lib/components/app/chat/ChatMessages/ChatMessageContinueRequest.svelte

@@ -0,0 +1,30 @@
+<script lang="ts">
+	import { RotateCw } from '@lucide/svelte';
+	import { Button } from '$lib/components/ui/button';
+	import ChatMessageActionCard from './ChatMessageActionCard.svelte';
+
+	interface Props {
+		onDecision: (shouldContinue: boolean) => void;
+	}
+
+	let { onDecision }: Props = $props();
+</script>
+
+<ChatMessageActionCard icon={RotateCw}>
+	{#snippet message()}
+		Agentic turn limit reached. Continue?
+	{/snippet}
+
+	{#snippet actions()}
+		<Button size="sm" onclick={() => onDecision(true)}>Continue</Button>
+
+		<Button
+			variant="destructive"
+			size="sm"
+			class="text-destructive hover:text-destructive"
+			onclick={() => onDecision(false)}
+		>
+			Stop
+		</Button>
+	{/snippet}
+</ChatMessageActionCard>

tools/server/webui/src/lib/components/app/chat/ChatMessages/ChatMessagePermissionRequest.svelte

@@ -0,0 +1,88 @@
+<script lang="ts">
+	import { ChevronDown, ShieldQuestion } from '@lucide/svelte';
+	import { Button } from '$lib/components/ui/button';
+	import * as ButtonGroup from '$lib/components/ui/button-group';
+	import * as DropdownMenu from '$lib/components/ui/dropdown-menu';
+	import { ToolSource, ToolPermissionDecision } from '$lib/enums';
+	import { TOOL_SERVER_LABELS } from '$lib/constants';
+	import { toolsStore } from '$lib/stores/tools.svelte';
+	import ChatMessageActionCard from './ChatMessageActionCard.svelte';
+
+	interface Props {
+		toolName: string;
+		serverLabel: string;
+		onDecision: (decision: ToolPermissionDecision) => void;
+	}
+
+	let { toolName, serverLabel, onDecision }: Props = $props();
+</script>
+
+<ChatMessageActionCard icon={ShieldQuestion}>
+	{#snippet message()}
+		Allow use of
+
+		<span class="font-semibold">{toolName}</span>
+
+		{#if serverLabel}
+			from <span class="font-semibold">{serverLabel}</span>
+		{/if}
+
+		?
+	{/snippet}
+
+	{#snippet actions()}
+		<DropdownMenu.Root>
+			<ButtonGroup.Root
+				class="overflow-hidden rounded-md bg-foreground text-white shadow-sm dark:bg-secondary dark:text-foreground"
+			>
+				<Button
+					class="rounded-none! shadow-none!"
+					size="sm"
+					onclick={() => onDecision(ToolPermissionDecision.ONCE)}
+				>
+					Allow once
+				</Button>
+
+				<ButtonGroup.Separator />
+
+				<DropdownMenu.Trigger>
+					<Button size="sm" class="rounded-none! !ps-2 shadow-none!">
+						<ChevronDown class="h-3.5 w-3.5" />
+					</Button>
+				</DropdownMenu.Trigger>
+			</ButtonGroup.Root>
+
+			<DropdownMenu.Content align="start" class="min-w-[8rem]">
+				<DropdownMenu.Item onclick={() => onDecision(ToolPermissionDecision.ALWAYS)}>
+					Always allow <pre>{toolName}</pre>
+					tool
+				</DropdownMenu.Item>
+				{#if serverLabel}
+					<DropdownMenu.Item onclick={() => onDecision(ToolPermissionDecision.ALWAYS_SERVER)}>
+						Always allow all tools from {serverLabel}
+					</DropdownMenu.Item>
+				{:else}
+					{@const source = toolsStore.getToolSource(toolName)}
+					{@const providerName =
+						source === ToolSource.BUILTIN
+							? TOOL_SERVER_LABELS[ToolSource.BUILTIN]
+							: source === ToolSource.CUSTOM
+								? TOOL_SERVER_LABELS[ToolSource.CUSTOM]
+								: 'MCP Tools'}
+					<DropdownMenu.Item onclick={() => onDecision(ToolPermissionDecision.ALWAYS_SERVER)}>
+						Approve all tools from {providerName}
+					</DropdownMenu.Item>
+				{/if}
+			</DropdownMenu.Content>
+		</DropdownMenu.Root>
+
+		<Button
+			variant="destructive"
+			size="sm"
+			class="text-destructive hover:text-destructive"
+			onclick={() => onDecision(ToolPermissionDecision.DENY)}
+		>
+			Deny
+		</Button>
+	{/snippet}
+</ChatMessageActionCard>

tools/server/webui/src/lib/components/app/chat/ChatMessages/ChatMessageUser.svelte

@@ -1,11 +1,9 @@
 <script lang="ts">
-	import { Card } from '$lib/components/ui/card';
-	import { ChatAttachmentsList, MarkdownContent } from '$lib/components/app';
 	import { getMessageEditContext } from '$lib/contexts';
-	import { config } from '$lib/stores/settings.svelte';
 	import ChatMessageActions from './ChatMessageActions.svelte';
 	import ChatMessageEditForm from './ChatMessageEditForm.svelte';
 	import { MessageRole } from '$lib/enums';
+	import ChatMessageUserBubble from './ChatMessageUserBubble.svelte';
 
 	interface Props {
 		class?: string;
@@ -44,34 +42,6 @@
 
 	// Get contexts
 	const editCtx = getMessageEditContext();
-
-	let isMultiline = $state(false);
-	let messageElement: HTMLElement | undefined = $state();
-	const currentConfig = config();
-
-	$effect(() => {
-		if (!messageElement || !message.content.trim()) return;
-
-		if (message.content.includes('\n')) {
-			isMultiline = true;
-			return;
-		}
-
-		const resizeObserver = new ResizeObserver((entries) => {
-			for (const entry of entries) {
-				const element = entry.target as HTMLElement;
-				const estimatedSingleLineHeight = 24; // Typical line height for text-md
-
-				isMultiline = element.offsetHeight > estimatedSingleLineHeight * 1.5;
-			}
-		});
-
-		resizeObserver.observe(messageElement);
-
-		return () => {
-			resizeObserver.disconnect();
-		};
-	});
 </script>
 
 <div
@@ -82,29 +52,11 @@
 	{#if editCtx.isEditing}
 		<ChatMessageEditForm />
 	{:else}
-		{#if message.extra && message.extra.length > 0}
-			<div class="mb-2 max-w-[80%]">
-				<ChatAttachmentsList attachments={message.extra} readonly imageHeight="h-80" />
-			</div>
-		{/if}
-
-		{#if message.content.trim()}
-			<Card
-				class="max-w-[80%] overflow-y-auto rounded-[1.125rem] border-none bg-primary/5 px-3.75 py-1.5 text-foreground backdrop-blur-md data-[multiline]:py-2.5 dark:bg-primary/15"
-				data-multiline={isMultiline ? '' : undefined}
-				style="max-height: var(--max-message-height); overflow-wrap: anywhere; word-break: break-word;"
-			>
-				{#if currentConfig.renderUserContentAsMarkdown}
-					<div bind:this={messageElement}>
-						<MarkdownContent class="markdown-user-content -my-4" content={message.content} />
-					</div>
-				{:else}
-					<span bind:this={messageElement} class="text-md whitespace-pre-wrap">
-						{message.content}
-					</span>
-				{/if}
-			</Card>
-		{/if}
+		<ChatMessageUserBubble
+			content={message.content}
+			attachments={message.extra}
+			renderMarkdown={true}
+		/>
 
 		{#if message.timestamp}
 			<div class="max-w-[80%]">

tools/server/webui/src/lib/components/app/chat/ChatMessages/ChatMessageUserBubble.svelte

@@ -0,0 +1,76 @@
+<script lang="ts">
+	import { Card } from '$lib/components/ui/card';
+	import { ChatAttachmentsList, MarkdownContent } from '$lib/components/app';
+	import { config } from '$lib/stores/settings.svelte';
+	import type { DatabaseMessageExtra } from '$lib/types/database';
+
+	interface Props {
+		content: string;
+		attachments?: DatabaseMessageExtra[];
+		renderMarkdown?: boolean;
+		textColorClass?: string;
+		cardBgClass?: string;
+		maxHeightStyle?: string;
+	}
+
+	let {
+		content,
+		attachments = [],
+		renderMarkdown = false,
+		textColorClass = 'text-foreground',
+		cardBgClass = 'dark:bg-primary/15',
+		maxHeightStyle = 'max-height: var(--max-message-height);'
+	}: Props = $props();
+
+	let isMultiline = $state(false);
+	let messageElement: HTMLElement | undefined = $state();
+	const currentConfig = config();
+
+	$effect(() => {
+		if (!messageElement || !content.trim()) return;
+
+		if (content.includes('\n')) {
+			isMultiline = true;
+			return;
+		}
+
+		const resizeObserver = new ResizeObserver((entries) => {
+			for (const entry of entries) {
+				const element = entry.target as HTMLElement;
+				const estimatedSingleLineHeight = 24; // Typical line height for text-md
+
+				isMultiline = element.offsetHeight > estimatedSingleLineHeight * 1.5;
+			}
+		});
+
+		resizeObserver.observe(messageElement);
+
+		return () => {
+			resizeObserver.disconnect();
+		};
+	});
+</script>
+
+{#if attachments && attachments.length > 0}
+	<div class="mb-2 max-w-[80%]">
+		<ChatAttachmentsList {attachments} readonly imageHeight="h-40" />
+	</div>
+{/if}
+
+{#if content.trim()}
+	<Card
+		class="max-w-[80%] overflow-y-auto rounded-[1.125rem] border-none bg-primary/5 px-3.75 py-1.5 {textColorClass} backdrop-blur-md data-[multiline]:py-2.5 {cardBgClass}"
+		data-multiline={isMultiline ? '' : undefined}
+		style="{maxHeightStyle} overflow-wrap: anywhere; word-break: break-word;"
+	>
+		{#if renderMarkdown && currentConfig.renderUserContentAsMarkdown}
+			<div bind:this={messageElement}>
+				<MarkdownContent class="markdown-user-content -my-4" {content} />
+			</div>
+		{:else}
+			<span bind:this={messageElement} class="text-md whitespace-pre-wrap">
+				{content}
+			</span>
+		{/if}
+	</Card>
+{/if}

tools/server/webui/src/lib/components/app/chat/ChatMessages/ChatMessageUserPending.svelte

@@ -0,0 +1,71 @@
+<script lang="ts">
+	import { ActionIcon } from '$lib/components/app';
+	import ChatMessageEditForm from './ChatMessageEditForm.svelte';
+	import { fadeInView } from '$lib/actions/fade-in-view.svelte';
+	import { ArrowUp, Edit, Trash2 } from '@lucide/svelte';
+	import { getProcessingInfoContext } from '$lib/contexts';
+	import { useMessageEditContext } from '$lib/hooks/use-message-edit-context.svelte';
+	import ChatMessageUserBubble from './ChatMessageUserBubble.svelte';
+
+	interface Props {
+		class?: string;
+		content: string;
+		extras?: DatabaseMessageExtra[];
+		onSendImmediately: () => void;
+		onEdit: (newContent: string, extras?: DatabaseMessageExtra[]) => void;
+		onDelete: () => void;
+	}
+
+	let {
+		class: className = '',
+		content,
+		extras = [],
+		onSendImmediately,
+		onEdit,
+		onDelete
+	}: Props = $props();
+
+	const processingInfoCtx = getProcessingInfoContext();
+	let showProcessingInfo = $derived(processingInfoCtx.showProcessingInfo);
+
+	const editCtx = useMessageEditContext({
+		getContent: () => content,
+		getExtras: () => extras,
+		onSave: (content, extras) => onEdit(content, extras)
+	});
+</script>
+
+<div
+	use:fadeInView
+	aria-label="Pending user message"
+	class="group flex flex-col items-end gap-3 transition-opacity hover:opacity-80 md:gap-2 {className} sticky {showProcessingInfo
+		? 'bottom-44'
+		: 'bottom-32'}"
+	role="group"
+>
+	{#if editCtx.isEditing}
+		<ChatMessageEditForm />
+	{:else}
+		<ChatMessageUserBubble
+			{content}
+			attachments={extras}
+			textColorClass="text-muted-foreground"
+			cardBgClass="dark:bg-primary/8"
+			maxHeightStyle="overflow-wrap: anywhere; word-break: break-word;"
+		/>
+
+		<div class="max-w-[80%]">
+			<div class="relative flex h-6 items-center justify-between">
+				<div class="right-0 flex items-center gap-2 opacity-100 transition-opacity">
+					<div
+						class="pointer-events-auto inset-0 flex items-center gap-1 opacity-0 transition-all duration-150 group-hover:opacity-100"
+					>
+						<ActionIcon icon={Edit} tooltip="Edit" onclick={editCtx.handleEdit} />
+						<ActionIcon icon={Trash2} tooltip="Delete" onclick={onDelete} />
+						<ActionIcon icon={ArrowUp} tooltip="Send immediately" onclick={onSendImmediately} />
+					</div>
+				</div>
+			</div>
+		</div>
+	{/if}
+</div>

tools/server/webui/src/lib/components/app/chat/ChatMessages/ChatMessages.svelte

@@ -1,11 +1,23 @@
 <script lang="ts">
-	import { fadeInView } from '$lib/actions/fade-in-view.svelte';
 	import { ChatMessage } from '$lib/components/app';
+	import ChatMessageUserPending from './ChatMessageUserPending.svelte';
 	import { setChatActionsContext } from '$lib/contexts';
 	import { MessageRole } from '$lib/enums';
 	import { chatStore } from '$lib/stores/chat.svelte';
+	import {
+		chatPendingMessageContent,
+		chatPendingMessageExtras,
+		chatClearPendingMessage,
+		chatInjectPendingMessage
+	} from '$lib/stores/chat.svelte';
 	import { conversationsStore, activeConversation } from '$lib/stores/conversations.svelte';
 	import { config } from '$lib/stores/settings.svelte';
+	import {
+		agenticPendingSteeringMessageContent,
+		agenticPendingSteeringMessageExtras,
+		agenticClearSteeringMessage,
+		agenticInjectSteeringMessage
+	} from '$lib/stores/agentic.svelte';
 	import {
 		copyToClipboard,
 		formatMessageForClipboard,
@@ -14,12 +26,11 @@
 	} from '$lib/utils';
 
 	interface Props {
-		class?: string;
 		messages?: DatabaseMessage[];
 		onUserAction?: () => void;
 	}
 
-	let { class: className, messages = [], onUserAction }: Props = $props();
+	let { messages = [], onUserAction }: Props = $props();
 
 	let allConversationMessages = $state<DatabaseMessage[]>([]);
 	const currentConfig = config();
@@ -196,19 +207,42 @@
 	});
 </script>
 
-<div
-	class="flex h-full flex-col space-y-10 pt-24 {className}"
-	style="height: auto; min-height: calc(100dvh - 14rem);"
->
-	{#each displayMessages as { message, toolMessages, isLastAssistantMessage, siblingInfo } (message.id)}
-		<div use:fadeInView>
-			<ChatMessage
-				class="mx-auto w-full max-w-[48rem]"
-				{message}
-				{toolMessages}
-				{isLastAssistantMessage}
-				{siblingInfo}
-			/>
-		</div>
-	{/each}
-</div>
+{#each displayMessages as { message, toolMessages, isLastAssistantMessage, siblingInfo } (message.id)}
+	<ChatMessage
+		class="mx-auto mt-12 w-full max-w-[48rem]"
+		{message}
+		{toolMessages}
+		{isLastAssistantMessage}
+		{siblingInfo}
+	/>
+{/each}
+
+{#if activeConversation() && agenticPendingSteeringMessageContent(activeConversation()!.id)}
+	{@const convId = activeConversation()!.id}
+	{@const pendingContent = agenticPendingSteeringMessageContent(convId)}
+
+	{#if pendingContent}
+		<ChatMessageUserPending
+			class="mx-auto mt-12 w-full max-w-[48rem]"
+			content={pendingContent}
+			extras={agenticPendingSteeringMessageExtras(convId)}
+			onSendImmediately={() => chatStore.abortCurrentFlow(convId)}
+			onEdit={(newContent, extras) => agenticInjectSteeringMessage(convId, newContent, extras)}
+			onDelete={() => agenticClearSteeringMessage(convId)}
+		/>
+	{/if}
+{:else if activeConversation() && chatPendingMessageContent(activeConversation()!.id)}
+	{@const convId = activeConversation()!.id}
+	{@const pendingContent = chatPendingMessageContent(convId)}
+
+	{#if pendingContent}
+		<ChatMessageUserPending
+			class="mx-auto mt-12 w-full max-w-[48rem]"
+			content={pendingContent}
+			extras={chatPendingMessageExtras(convId)}
+			onSendImmediately={() => chatStore.abortCurrentFlow(convId)}
+			onEdit={(newContent, extras) => chatInjectPendingMessage(convId, newContent, extras)}
+			onDelete={() => chatClearPendingMessage(convId)}
+		/>
+	{/if}
+{/if}

tools/server/webui/src/lib/components/app/chat/ChatScreen/ChatScreen.svelte

@@ -2,7 +2,6 @@
 	import { afterNavigate } from '$app/navigation';
 	import {
 		ChatScreenForm,
-		ChatScreenHeader,
 		ChatMessages,
 		ChatScreenProcessingInfo,
 		DialogEmptyFileAlert,
@@ -12,15 +11,16 @@
 	} from '$lib/components/app';
 	import * as Alert from '$lib/components/ui/alert';
 	import * as AlertDialog from '$lib/components/ui/alert-dialog';
-	import { KeyboardKey } from '$lib/enums';
 	import { createAutoScrollController } from '$lib/hooks/use-auto-scroll.svelte';
+	import { useKeyboardShortcuts } from '$lib/hooks/use-keyboard-shortcuts.svelte';
 	import {
 		chatStore,
 		errorDialog,
 		isLoading,
 		isChatStreaming,
 		isEditing,
-		getAddFilesHandler
+		getAddFilesHandler,
+		activeProcessingState
 	} from '$lib/stores/chat.svelte';
 	import {
 		conversationsStore,
@@ -34,9 +34,11 @@
 	import { parseFilesToMessageExtras, processFilesToChatUploaded } from '$lib/utils/browser-only';
 	import { ErrorDialogType } from '$lib/enums';
 	import { onMount } from 'svelte';
-	import { fade, fly, slide } from 'svelte/transition';
+	import { fadeInView } from '$lib/actions/fade-in-view.svelte';
 	import { Trash2, AlertTriangle, RefreshCw } from '@lucide/svelte';
 	import ChatScreenDragOverlay from './ChatScreenDragOverlay.svelte';
+	import { page } from '$app/state';
+	import { setProcessingInfoContext } from '$lib/contexts';
 
 	let { showCenteredEmpty = false } = $props();
 
@@ -79,6 +81,18 @@
 
 	let isCurrentConversationLoading = $derived(isLoading() || isChatStreaming());
 
+	let showProcessingInfo = $derived(
+		isCurrentConversationLoading ||
+			(config().keepStatsVisible && !!page.params.id) ||
+			activeProcessingState() !== null
+	);
+
+	setProcessingInfoContext({
+		get showProcessingInfo() {
+			return showProcessingInfo;
+		}
+	});
+
 	let isRouter = $derived(isRouterMode());
 
 	let conversationModel = $derived(
@@ -208,20 +222,13 @@
 		processFiles(files);
 	}
 
-	function handleKeydown(event: KeyboardEvent) {
-		const isCtrlOrCmd = event.ctrlKey || event.metaKey;
-
-		if (
-			isCtrlOrCmd &&
-			event.shiftKey &&
-			(event.key === KeyboardKey.D_LOWER || event.key === KeyboardKey.D_UPPER)
-		) {
-			event.preventDefault();
+	const { handleKeydown } = useKeyboardShortcuts({
+		deleteActiveConversation: () => {
 			if (activeConversation()) {
 				showDeleteDialog = true;
 			}
 		}
-	}
+	});
 
 	async function handleSystemPromptAdd(draft: { message: string; files: ChatUploadedFile[] }) {
 		if (draft.message || draft.files.length > 0) {
@@ -342,9 +349,9 @@
 
 <svelte:window onkeydown={handleKeydown} />
 
-<ChatScreenHeader />
-
-{#if !isEmpty}
+{#if isServerLoading}
+	<ServerLoadingSplash />
+{:else}
 	<div
 		bind:this={chatScrollContainer}
 		aria-label="Chat interface with file drop zone"
@@ -356,26 +363,42 @@
 		onscroll={handleScroll}
 		role="main"
 	>
-		<div class="flex flex-col">
-			<ChatMessages
-				class="mb-16 md:mb-24"
-				messages={activeMessages()}
-				onUserAction={() => {
-					autoScroll.enable();
-					autoScroll.scrollToBottom();
-				}}
-			/>
+		<div class="flex grow flex-col pt-14">
+			{#if !isEmpty}
+				<ChatMessages
+					messages={activeMessages()}
+					onUserAction={() => {
+						autoScroll.enable();
+						autoScroll.scrollToBottom();
+					}}
+				/>
+			{/if}
 
 			<div
-				class="pointer-events-none sticky right-0 bottom-4 left-0 mt-auto"
-				in:slide={{ duration: 150, axis: 'y' }}
+				class="pointer-events-none {isEmpty
+					? 'absolute bottom-[calc(50dvh-7rem)]'
+					: 'sticky bottom-4'} right-4 left-4 mt-auto pt-16 transition-all duration-200"
 			>
-				<ChatScreenProcessingInfo />
+				{#if isEmpty}
+					<div class="mb-8 px-4 text-center" use:fadeInView={{ duration: 300 }}>
+						<h1 class="mb-2 text-2xl font-semibold tracking-tight md:text-3xl">Hello there</h1>
+
+						<p class="text-muted-foreground md:text-lg">
+							{serverStore.props?.modalities?.audio
+								? 'Record audio, type a message '
+								: 'Type a message'} or upload files to get started
+						</p>
+					</div>
+				{/if}
+
+				{#if page.params.id}
+					<ChatScreenProcessingInfo />
+				{/if}
 
 				{#if hasPropsError}
 					<div
 						class="pointer-events-auto mx-auto mb-4 max-w-[48rem] px-1"
-						in:fly={{ y: 10, duration: 250 }}
+						use:fadeInView={{ y: 10, duration: 250 }}
 					>
 						<Alert.Root variant="destructive">
 							<AlertTriangle class="h-4 w-4" />
@@ -412,69 +435,6 @@
 			</div>
 		</div>
 	</div>
-{:else if isServerLoading}
-	<!-- Server Loading State -->
-	<ServerLoadingSplash />
-{:else}
-	<div
-		aria-label="Welcome screen with file drop zone"
-		class="flex h-full items-center justify-center"
-		ondragenter={handleDragEnter}
-		ondragleave={handleDragLeave}
-		ondragover={handleDragOver}
-		ondrop={handleDrop}
-		role="main"
-	>
-		<div class="w-full max-w-[48rem] px-4">
-			<div class="mb-10 text-center" in:fade={{ duration: 300 }}>
-				<h1 class="mb-2 text-2xl font-semibold tracking-tight md:text-3xl">llama.cpp UI (KoboldCpp Backend)</h1>
-
-				<p class="text-muted-foreground md:text-lg">
-					{serverStore.props?.modalities?.audio
-						? 'Record audio, type a message '
-						: 'Type a message'} or upload files to get started
-				</p>
-			</div>
-
-			{#if hasPropsError}
-				<div class="mb-4" in:fly={{ y: 10, duration: 250 }}>
-					<Alert.Root variant="destructive">
-						<AlertTriangle class="h-4 w-4" />
-
-						<Alert.Title class="flex items-center justify-between">
-							<span>Server unavailable</span>
-
-							<button
-								onclick={() => serverStore.fetch()}
-								disabled={isServerLoading}
-								class="flex items-center gap-1.5 rounded-lg bg-destructive/20 px-2 py-1 text-xs font-medium hover:bg-destructive/30 disabled:opacity-50"
-							>
-								<RefreshCw class="h-3 w-3 {isServerLoading ? 'animate-spin' : ''}" />
-								{isServerLoading ? 'Retrying...' : 'Retry'}
-							</button>
-						</Alert.Title>
-
-						<Alert.Description>{serverError()}</Alert.Description>
-					</Alert.Root>
-				</div>
-			{/if}
-
-			<div in:fly={{ y: 10, duration: 250, delay: hasPropsError ? 0 : 300 }}>
-				<ChatScreenForm
-					disabled={hasPropsError}
-					{initialMessage}
-					isLoading={isCurrentConversationLoading}
-					onFileRemove={handleFileRemove}
-					onFileUpload={handleFileUpload}
-					onSend={handleSendMessage}
-					onStop={() => chatStore.stopGeneration()}
-					onSystemPromptAdd={handleSystemPromptAdd}
-					showHelperText
-					bind:uploadedFiles
-				/>
-			</div>
-		</div>
-	</div>
 {/if}
 
 <!-- File Upload Error Alert Dialog -->
@@ -575,21 +535,3 @@
 	open={Boolean(activeErrorDialog)}
 	type={activeErrorDialog?.type ?? ErrorDialogType.SERVER}
 />
-
-<style>
-	.conversation-chat-form {
-		position: relative;
-
-		&::after {
-			content: '';
-			position: absolute;
-			bottom: 0;
-			z-index: -1;
-			left: 0;
-			right: 0;
-			width: 100%;
-			height: 2.375rem;
-			background-color: var(--background);
-		}
-	}
-</style>

tools/server/webui/src/lib/components/app/chat/ChatScreen/ChatScreenForm.svelte

@@ -1,7 +1,9 @@
 <script lang="ts">
 	import { afterNavigate } from '$app/navigation';
+	import { page } from '$app/state';
 	import { ChatFormHelperText, ChatForm } from '$lib/components/app';
 	import { onMount } from 'svelte';
+	import { useDraftMessages } from '$lib/hooks/use-draft-messages.svelte';
 
 	interface Props {
 		class?: string;
@@ -32,11 +34,20 @@
 	}: Props = $props();
 
 	let chatFormRef: ChatForm | undefined = $state(undefined);
+	let chatId = $derived(page.params.id as string | undefined);
 	let message = $derived(initialMessage);
 	let previousIsLoading = $derived(isLoading);
 	let previousInitialMessage = $derived(initialMessage);
 
-	// Sync message when initialMessage prop changes (e.g., after draft restoration)
+	const { clearDraft } = useDraftMessages({
+		getChatId: () => chatId,
+		getMessage: () => message,
+		getFiles: () => uploadedFiles,
+		setMessage: (m) => (message = m),
+		setFiles: (f) => (uploadedFiles = f),
+		getInitialMessage: () => initialMessage
+	});
+
 	$effect(() => {
 		if (initialMessage !== previousInitialMessage) {
 			message = initialMessage;
@@ -51,12 +62,7 @@
 	let hasLoadingAttachments = $derived(uploadedFiles.some((f) => f.isLoading));
 
 	async function handleSubmit() {
-		if (
-			(!message.trim() && uploadedFiles.length === 0) ||
-			disabled ||
-			isLoading ||
-			hasLoadingAttachments
-		)
+		if ((!message.trim() && uploadedFiles.length === 0) || disabled || hasLoadingAttachments)
 			return;
 
 		if (!chatFormRef?.checkModelSelected()) return;
@@ -66,6 +72,7 @@
 
 		message = '';
 		uploadedFiles = [];
+		clearDraft();
 
 		chatFormRef?.resetTextareaHeight();
 
@@ -89,8 +96,10 @@
 		setTimeout(() => chatFormRef?.focus(), 10);
 	});
 
-	afterNavigate(() => {
-		setTimeout(() => chatFormRef?.focus(), 10);
+	afterNavigate((navigation) => {
+		if (navigation?.from != null) {
+			setTimeout(() => chatFormRef?.focus(), 10);
+		}
 	});
 
 	$effect(() => {

tools/server/webui/src/lib/components/app/chat/ChatScreen/ChatScreenHeader.svelte

@@ -1,26 +0,0 @@
-<script lang="ts">
-	import { Settings } from '@lucide/svelte';
-	import { Button } from '$lib/components/ui/button';
-	import { useSidebar } from '$lib/components/ui/sidebar';
-	import { getChatSettingsDialogContext } from '$lib/contexts';
-
-	const sidebar = useSidebar();
-	const chatSettingsDialog = getChatSettingsDialogContext();
-</script>
-
-<header
-	class="pointer-events-none fixed top-0 right-0 left-0 z-50 flex items-center justify-end p-2 duration-200 ease-linear md:p-4 {sidebar.open
-		? 'md:left-[var(--sidebar-width)]'
-		: ''}"
->
-	<div class="pointer-events-auto flex items-center space-x-2">
-		<Button
-			variant="ghost"
-			size="icon-lg"
-			onclick={() => chatSettingsDialog.open()}
-			class="rounded-full backdrop-blur-lg"
-		>
-			<Settings class="h-4 w-4" />
-		</Button>
-	</div>
-</header>

tools/server/webui/src/lib/components/app/chat/ChatScreen/ChatScreenProcessingInfo.svelte

@@ -5,18 +5,17 @@
 	import { chatStore, isLoading, isChatStreaming } from '$lib/stores/chat.svelte';
 	import { activeMessages, activeConversation } from '$lib/stores/conversations.svelte';
 	import { config } from '$lib/stores/settings.svelte';
+	import { getProcessingInfoContext } from '$lib/contexts';
 
 	const processingState = useProcessingState();
+	const processingInfoCtx = getProcessingInfoContext();
+
+	let showProcessingInfo = $derived(processingInfoCtx.showProcessingInfo);
 
 	let isCurrentConversationLoading = $derived(isLoading());
 	let isStreaming = $derived(isChatStreaming());
-	let hasProcessingData = $derived(processingState.processingState !== null);
 	let processingDetails = $derived(processingState.getTechnicalDetails());
 
-	let showProcessingInfo = $derived(
-		isCurrentConversationLoading || isStreaming || config().keepStatsVisible || hasProcessingData
-	);
-
 	$effect(() => {
 		const conversation = activeConversation();
 

tools/server/webui/src/lib/components/app/chat/ChatSettings/ChatSettings.svelte

@@ -7,16 +7,10 @@
 		Monitor,
 		ChevronLeft,
 		ChevronRight,
-		Database
+		ListRestart,
+		Sliders
 	} from '@lucide/svelte';
-	import {
-		ChatSettingsFooter,
-		ChatSettingsImportExportTab,
-		ChatSettingsFields,
-		McpLogo,
-		McpServersSettings
-	} from '$lib/components/app';
-	import { ScrollArea } from '$lib/components/ui/scroll-area';
+	import { ChatSettingsFooter, ChatSettingsFields } from '$lib/components/app';
 	import { config, settingsStore } from '$lib/stores/settings.svelte';
 	import {
 		SETTINGS_SECTION_TITLES,
@@ -29,14 +23,16 @@
 	import { setMode } from 'mode-watcher';
 	import { ColorMode } from '$lib/enums/ui';
 	import { SettingsFieldType } from '$lib/enums/settings';
+	import { fade } from 'svelte/transition';
 	import type { Component } from 'svelte';
 
 	interface Props {
+		class?: string;
 		onSave?: () => void;
 		initialSection?: SettingsSectionTitle;
 	}
 
-	let { onSave, initialSection }: Props = $props();
+	let { class: className, onSave, initialSection }: Props = $props();
 
 	const settingSections: Array<{
 		fields: SettingsFieldConfig[];
@@ -45,7 +41,7 @@
 	}> = [
 		{
 			title: SETTINGS_SECTION_TITLES.GENERAL,
-			icon: Settings,
+			icon: Sliders,
 			fields: [
 				{
 					key: SETTINGS_KEYS.THEME,
@@ -111,6 +107,11 @@
 					label: 'Show thought in progress',
 					type: SettingsFieldType.CHECKBOX
 				},
+				{
+					key: SETTINGS_KEYS.SHOW_TOOL_CALL_IN_PROGRESS,
+					label: 'Show tool call in progress',
+					type: SettingsFieldType.CHECKBOX
+				},
 				{
 					key: SETTINGS_KEYS.KEEP_STATS_VISIBLE,
 					label: 'Keep stats visible after generation',
@@ -143,13 +144,13 @@
 					type: SettingsFieldType.CHECKBOX
 				},
 				{
-					key: SETTINGS_KEYS.AUTO_SHOW_SIDEBAR_ON_NEW_CHAT,
-					label: 'Auto-show sidebar on new chat',
+					key: SETTINGS_KEYS.SHOW_RAW_MODEL_NAMES,
+					label: 'Show raw model names',
 					type: SettingsFieldType.CHECKBOX
 				},
 				{
-					key: SETTINGS_KEYS.SHOW_RAW_MODEL_NAMES,
-					label: 'Show raw model names',
+					key: SETTINGS_KEYS.ALWAYS_SHOW_AGENTIC_TURNS,
+					label: 'Always show agentic turns in conversation',
 					type: SettingsFieldType.CHECKBOX
 				}
 			]
@@ -267,33 +268,18 @@
 			]
 		},
 		{
-			title: SETTINGS_SECTION_TITLES.IMPORT_EXPORT,
-			icon: Database,
-			fields: []
-		},
-		{
-			title: SETTINGS_SECTION_TITLES.MCP,
-			icon: McpLogo,
+			title: SETTINGS_SECTION_TITLES.AGENTIC,
+			icon: ListRestart,
 			fields: [
 				{
 					key: SETTINGS_KEYS.AGENTIC_MAX_TURNS,
-					label: 'Agentic loop max turns',
+					label: 'Agentic turns',
 					type: SettingsFieldType.INPUT
 				},
-				{
-					key: SETTINGS_KEYS.ALWAYS_SHOW_AGENTIC_TURNS,
-					label: 'Always show agentic turns in conversation',
-					type: SettingsFieldType.CHECKBOX
-				},
 				{
 					key: SETTINGS_KEYS.AGENTIC_MAX_TOOL_PREVIEW_LINES,
 					label: 'Max lines per tool preview',
 					type: SettingsFieldType.INPUT
-				},
-				{
-					key: SETTINGS_KEYS.SHOW_TOOL_CALL_IN_PROGRESS,
-					label: 'Show tool call in progress',
-					type: SettingsFieldType.CHECKBOX
 				}
 			]
 		},
@@ -457,119 +443,116 @@
 	});
 </script>
 
-<div class="flex h-full flex-col overflow-hidden md:flex-row">
-	<!-- Desktop Sidebar -->
-	<div class="hidden w-64 border-r border-border/30 p-6 md:block">
-		<nav class="space-y-1 py-2">
-			{#each settingSections as section (section.title)}
-				<button
-					class="flex w-full cursor-pointer items-center gap-3 rounded-lg px-3 py-2 text-left text-sm transition-colors hover:bg-accent {activeSection ===
-					section.title
-						? 'bg-accent text-accent-foreground'
-						: 'text-muted-foreground'}"
-					onclick={() => (activeSection = section.title)}
-				>
-					<section.icon class="h-4 w-4" />
-
-					<span class="ml-2">{section.title}</span>
-				</button>
-			{/each}
-		</nav>
-	</div>
-
-	<!-- Mobile Header with Horizontal Scrollable Menu -->
-	<div class="flex flex-col pt-6 md:hidden">
-		<div class="border-b border-border/30 pt-4 md:py-4">
-			<!-- Horizontal Scrollable Category Menu with Navigation -->
-			<div class="relative flex items-center" style="scroll-padding: 1rem;">
-				<button
-					class="absolute left-2 z-10 flex h-6 w-6 items-center justify-center rounded-full bg-muted shadow-md backdrop-blur-sm transition-opacity hover:bg-accent {canScrollLeft
-						? 'opacity-100'
-						: 'pointer-events-none opacity-0'}"
-					onclick={scrollLeft}
-					aria-label="Scroll left"
-				>
-					<ChevronLeft class="h-4 w-4" />
-				</button>
-
-				<div
-					class="scrollbar-hide overflow-x-auto py-2"
-					bind:this={scrollContainer}
-					onscroll={updateScrollButtons}
-				>
-					<div class="flex min-w-max gap-2">
-						{#each settingSections as section (section.title)}
-							<button
-								class="flex cursor-pointer items-center gap-2 rounded-lg px-3 py-2 text-sm whitespace-nowrap transition-colors first:ml-4 last:mr-4 hover:bg-accent {activeSection ===
-								section.title
-									? 'bg-accent text-accent-foreground'
-									: 'text-muted-foreground'}"
-								onclick={(e: MouseEvent) => {
-									activeSection = section.title;
-									scrollToCenter(e.currentTarget as HTMLElement);
-								}}
-							>
-								<section.icon class="h-4 w-4 flex-shrink-0" />
-								<span>{section.title}</span>
-							</button>
-						{/each}
-					</div>
-				</div>
-
-				<button
-					class="absolute right-2 z-10 flex h-6 w-6 items-center justify-center rounded-full bg-muted shadow-md backdrop-blur-sm transition-opacity hover:bg-accent {canScrollRight
-						? 'opacity-100'
-						: 'pointer-events-none opacity-0'}"
-					onclick={scrollRight}
-					aria-label="Scroll right"
-				>
-					<ChevronRight class="h-4 w-4" />
-				</button>
+<div class="flex h-full flex-col overflow-y-auto {className} w-full" in:fade={{ duration: 150 }}>
+	<div class="flex flex-1 flex-col gap-4 md:flex-row">
+		<!-- Desktop Sidebar -->
+		<div class="sticky top-0 hidden w-64 flex-col self-start bg-background pt-8 pb-4 md:flex">
+			<div class="flex items-center gap-2 pb-8">
+				<Settings class="h-6 w-6" />
+				<h1 class="text-2xl font-semibold">Settings</h1>
 			</div>
+			<nav class="space-y-1">
+				{#each settingSections as section (section.title)}
+					<button
+						class="flex w-full cursor-pointer items-center gap-3 rounded-lg px-3 py-2 text-left text-sm transition-colors hover:bg-accent {activeSection ===
+						section.title
+							? 'bg-accent text-accent-foreground'
+							: 'text-muted-foreground'}"
+						onclick={() => (activeSection = section.title)}
+					>
+						<section.icon class="h-4 w-4" />
+
+						<span class="ml-2">{section.title}</span>
+					</button>
+				{/each}
+			</nav>
 		</div>
-	</div>
 
-	<ScrollArea class="max-h-[calc(100dvh-13.5rem)] flex-1 md:max-h-[calc(100vh-13.5rem)]">
-		<div class="space-y-6 p-4 md:p-6">
-			<div class="grid">
-				<div class="mb-6 flex hidden items-center gap-2 border-b border-border/30 pb-6 md:flex">
-					<currentSection.icon class="h-5 w-5" />
+		<!-- Mobile Header with Horizontal Scrollable Menu -->
+		<div class="sticky top-0 z-10 flex flex-col bg-background md:hidden">
+			<div class="flex items-center gap-2 px-4 pt-4 pb-2 md:pt-6">
+				<Settings class="h-5 w-5 md:h-6 md:w-6" />
 
-					<h3 class="text-lg font-semibold">{currentSection.title}</h3>
-				</div>
+				<h1 class="text-xl font-semibold md:text-2xl">Settings</h1>
+			</div>
 
-				{#if currentSection.title === SETTINGS_SECTION_TITLES.IMPORT_EXPORT}
-					<ChatSettingsImportExportTab />
-				{:else if currentSection.title === SETTINGS_SECTION_TITLES.MCP}
-					<div class="space-y-6">
-						<ChatSettingsFields
-							fields={currentSection.fields}
-							{localConfig}
-							onConfigChange={handleConfigChange}
-							onThemeChange={handleThemeChange}
-						/>
+			<div class="border-b border-border/30 py-2">
+				<!-- Horizontal Scrollable Category Menu with Navigation -->
+				<div class="relative flex items-center" style="scroll-padding: 1rem;">
+					<button
+						class="absolute left-2 z-10 flex h-6 w-6 items-center justify-center rounded-full bg-muted shadow-md backdrop-blur-sm transition-opacity hover:bg-accent {canScrollLeft
+							? 'opacity-100'
+							: 'pointer-events-none opacity-0'}"
+						onclick={scrollLeft}
+						aria-label="Scroll left"
+					>
+						<ChevronLeft class="h-4 w-4" />
+					</button>
 
-						<div class="border-t border-border/30 pt-6">
-							<McpServersSettings />
+					<div
+						class="scrollbar-hide overflow-x-auto py-2"
+						bind:this={scrollContainer}
+						onscroll={updateScrollButtons}
+					>
+						<div class="flex min-w-max gap-2">
+							{#each settingSections as section (section.title)}
+								<button
+									class="flex cursor-pointer items-center gap-2 rounded-lg px-3 py-2 text-sm whitespace-nowrap transition-colors first:ml-4 last:mr-4 hover:bg-accent {activeSection ===
+									section.title
+										? 'bg-accent text-accent-foreground'
+										: 'text-muted-foreground'}"
+									onclick={(e: MouseEvent) => {
+										activeSection = section.title;
+										scrollToCenter(e.currentTarget as HTMLElement);
+									}}
+								>
+									<section.icon class="h-4 w-4 flex-shrink-0" />
+									<span>{section.title}</span>
+								</button>
+							{/each}
 						</div>
 					</div>
-				{:else}
-					<div class="space-y-6">
-						<ChatSettingsFields
-							fields={currentSection.fields}
-							{localConfig}
-							onConfigChange={handleConfigChange}
-							onThemeChange={handleThemeChange}
-						/>
-					</div>
-				{/if}
-			</div>
 
-			<div class="mt-8 border-t pt-6">
-				<p class="text-xs text-muted-foreground">Settings are saved in browser's localStorage</p>
+					<button
+						class="absolute right-2 z-10 flex h-6 w-6 items-center justify-center rounded-full bg-muted shadow-md backdrop-blur-sm transition-opacity hover:bg-accent {canScrollRight
+							? 'opacity-100'
+							: 'pointer-events-none opacity-0'}"
+						onclick={scrollRight}
+						aria-label="Scroll right"
+					>
+						<ChevronRight class="h-4 w-4" />
+					</button>
+				</div>
 			</div>
 		</div>
-	</ScrollArea>
-</div>
 
-<ChatSettingsFooter onReset={handleReset} onSave={handleSave} />
+		<div class="mx-auto max-w-3xl flex-1">
+			<div class="space-y-6 p-4 md:p-6 md:pt-28">
+				<div class="grid">
+					<div class="mb-6 flex hidden items-center gap-2 border-b border-border/30 pb-6 md:flex">
+						<currentSection.icon class="h-5 w-5" />
+
+						<h3 class="text-lg font-semibold">{currentSection.title}</h3>
+					</div>
+
+					{#if currentSection.fields}
+						<div class="space-y-6">
+							<ChatSettingsFields
+								fields={currentSection.fields}
+								{localConfig}
+								onConfigChange={handleConfigChange}
+								onThemeChange={handleThemeChange}
+							/>
+						</div>
+					{/if}
+				</div>
+
+				<div class="mt-8 border-t border-border/30 pt-6">
+					<p class="text-xs text-muted-foreground">Settings are saved in browser's localStorage</p>
+				</div>
+			</div>
+
+			<ChatSettingsFooter onReset={handleReset} onSave={handleSave} />
+		</div>
+	</div>
+</div>

tools/server/webui/src/lib/components/app/chat/ChatSettings/ChatSettingsFields.svelte

@@ -70,7 +70,7 @@
 				{/if}
 			</div>
 
-			<div class="relative w-full md:max-w-md">
+			<div class="relative w-full">
 				<Input
 					id={field.key}
 					value={currentValue}
@@ -117,7 +117,7 @@
 				value={String(localConfig[field.key] ?? '')}
 				onchange={(e) => onConfigChange(field.key, e.currentTarget.value)}
 				placeholder=""
-				class="min-h-[10rem] w-full md:max-w-2xl"
+				class="min-h-[10rem] w-full md:max-w-3xl"
 			/>
 
 			{#if field.help || SETTING_CONFIG_INFO[field.key]}
@@ -176,7 +176,7 @@
 					}
 				}}
 			>
-				<div class="relative w-full md:w-auto md:max-w-md">
+				<div class="relative w-full md:w-auto">
 					<Select.Trigger class="w-full">
 						<div class="flex items-center gap-2">
 							{#if selectedOption?.icon}

tools/server/webui/src/lib/components/app/chat/ChatSettings/ChatSettingsFooter.svelte

@@ -29,7 +29,7 @@
 	}
 </script>
 
-<div class="flex justify-between border-t border-border/30 p-6">
+<div class="sticky bottom-0 mx-auto mt-4 flex w-full justify-between p-6">
 	<div class="flex gap-2">
 		<Button variant="outline" onclick={handleResetClick}>
 			<RotateCcw class="h-3 w-3" />

tools/server/webui/src/lib/components/app/chat/ChatSettings/ChatSettingsToolsTab.svelte

@@ -0,0 +1,122 @@
+<script lang="ts">
+	import { ChevronDown, ChevronRight } from '@lucide/svelte';
+	import { Checkbox } from '$lib/components/ui/checkbox';
+	import * as Collapsible from '$lib/components/ui/collapsible';
+	import { TruncatedText } from '$lib/components/app';
+	import { toolsStore } from '$lib/stores/tools.svelte';
+	import { permissionsStore } from '$lib/stores/permissions.svelte';
+	import { mcpStore } from '$lib/stores/mcp.svelte';
+	import { ToolSource } from '$lib/enums';
+	import { SvelteSet } from 'svelte/reactivity';
+
+	let expandedGroups = new SvelteSet<string>();
+	let groups = $derived(toolsStore.toolGroups);
+
+	function getFavicon(group: { source: ToolSource; label: string }): string | null {
+		if (group.source !== ToolSource.MCP) return null;
+
+		for (const server of mcpStore.getServersSorted()) {
+			if (mcpStore.getServerLabel(server) === group.label) {
+				return mcpStore.getServerFavicon(server.id);
+			}
+		}
+
+		return null;
+	}
+
+	function toggleExpanded(label: string) {
+		if (expandedGroups.has(label)) {
+			expandedGroups.delete(label);
+		} else {
+			expandedGroups.add(label);
+		}
+	}
+</script>
+
+{#if groups.length === 0}
+	<div class="py-8 text-center text-sm text-muted-foreground">No tools available</div>
+{:else}
+	<div class="space-y-2">
+		{#each groups as group (group.label)}
+			{@const isExpanded = expandedGroups.has(group.label)}
+			{@const favicon = getFavicon(group)}
+
+			<Collapsible.Root open={isExpanded} onOpenChange={() => toggleExpanded(group.label)}>
+				<Collapsible.Trigger
+					class="flex w-full items-center gap-2 rounded-lg px-3 py-2 text-sm hover:bg-muted/50"
+				>
+					{#if isExpanded}
+						<ChevronDown class="h-3.5 w-3.5 shrink-0" />
+					{:else}
+						<ChevronRight class="h-3.5 w-3.5 shrink-0" />
+					{/if}
+
+					<span class="inline-flex min-w-0 items-center gap-1.5 font-medium">
+						{#if favicon}
+							<img
+								src={favicon}
+								alt=""
+								class="h-4 w-4 shrink-0 rounded-sm"
+								onerror={(e) => {
+									(e.currentTarget as HTMLImageElement).style.display = 'none';
+								}}
+							/>
+						{/if}
+
+						<span class="truncate">{group.label}</span>
+					</span>
+
+					<span class="ml-auto shrink-0 text-xs text-muted-foreground">
+						{group.tools.length} tool{group.tools.length !== 1 ? 's' : ''}
+					</span>
+				</Collapsible.Trigger>
+
+				<Collapsible.Content>
+					<div class="ml-4 border-l border-border/50 pl-2">
+						<!-- Header row -->
+						<div class="flex items-center gap-2 px-2 py-1 text-xs text-muted-foreground">
+							<span class="min-w-0 flex-1">Tool</span>
+							<span class="w-16 shrink-0 text-center">Enabled</span>
+							<span class="w-20 shrink-0 text-center">Always allow</span>
+						</div>
+
+						{#each group.tools as tool (tool.function.name)}
+							{@const toolName = tool.function.name}
+							{@const isEnabled = toolsStore.isToolEnabled(toolName)}
+							{@const permissionKey = toolsStore.getPermissionKey(toolName)}
+							{@const isAlwaysAllowed = permissionKey
+								? permissionsStore.hasTool(permissionKey)
+								: false}
+
+							<div class="flex items-center gap-2 rounded px-2 py-1.5 text-sm hover:bg-muted/50">
+								<TruncatedText text={toolName} class="min-w-0 flex-1 truncate" showTooltip={true} />
+
+								<div class="flex w-16 shrink-0 justify-center">
+									<Checkbox
+										checked={isEnabled}
+										onCheckedChange={() => toolsStore.toggleTool(toolName)}
+										class="h-4 w-4"
+									/>
+								</div>
+
+								<div class="flex w-20 shrink-0 justify-center">
+									<Checkbox
+										checked={isAlwaysAllowed}
+										onCheckedChange={() => {
+											if (isAlwaysAllowed) {
+												permissionsStore.revokeTool(permissionKey!);
+											} else {
+												permissionsStore.allowTool(permissionKey!);
+											}
+										}}
+										class="h-4 w-4"
+									/>
+								</div>
+							</div>
+						{/each}
+					</div>
+				</Collapsible.Content>
+			</Collapsible.Root>
+		{/each}
+	</div>
+{/if}

tools/server/webui/src/lib/components/app/chat/ChatSidebar/ChatSidebar.svelte

@@ -1,7 +1,8 @@
 <script lang="ts">
 	import { goto } from '$app/navigation';
 	import { page } from '$app/state';
-	import { Trash2, Pencil } from '@lucide/svelte';
+	import { Trash2, Pencil, X } from '@lucide/svelte';
+	import { Button } from '$lib/components/ui/button';
 	import { ChatSidebarConversationItem, DialogConfirmation } from '$lib/components/app';
 	import { Checkbox } from '$lib/components/ui/checkbox';
 	import Label from '$lib/components/ui/label/label.svelte';
@@ -16,6 +17,7 @@
 	import { chatStore } from '$lib/stores/chat.svelte';
 	import { getPreviewText } from '$lib/utils';
 	import ChatSidebarActions from './ChatSidebarActions.svelte';
+	import { APP_NAME } from '$lib/constants';
 
 	const sidebar = Sidebar.useSidebar();
 
@@ -32,10 +34,14 @@
 	);
 
 	let filteredConversations = $derived.by(() => {
-		if (searchQuery.trim().length > 0) {
-			return conversations().filter((conversation: { name: string }) =>
-				conversation.name.toLowerCase().includes(searchQuery.toLowerCase())
-			);
+		if (isSearchModeActive) {
+			if (searchQuery.trim().length > 0) {
+				return conversations().filter((conversation: { name: string }) =>
+					conversation.name.toLowerCase().includes(searchQuery.toLowerCase())
+				);
+			}
+
+			return [];
 		}
 
 		return conversations();
@@ -107,10 +113,31 @@
 		}
 	}
 
+	let chatSidebarActions: { activateSearch?: () => void } | undefined = $state();
+	let openedForSearch = $state(false);
+
 	export function activateSearchMode() {
-		isSearchModeActive = true;
+		if (!sidebar.open) {
+			openedForSearch = true;
+		}
+		chatSidebarActions?.activateSearch?.();
 	}
 
+	function handleSearchDeactivated() {
+		if (openedForSearch) {
+			openedForSearch = false;
+			sidebar.toggle();
+		}
+	}
+
+	$effect(() => {
+		if (!sidebar.open) {
+			isSearchModeActive = false;
+			searchQuery = '';
+			openedForSearch = false;
+		}
+	});
+
 	export function editActiveConversation() {
 		if (currentChatId) {
 			const activeConversation = filteredConversations.find((conv) => conv.id === currentChatId);
@@ -130,6 +157,7 @@
 			searchQuery = '';
 		}
 
+		handleMobileSidebarItemClick();
 		await goto(`#/chat/${id}`);
 	}
 
@@ -138,60 +166,79 @@
 	}
 </script>
 
-<ScrollArea class="h-[100vh]">
-	<Sidebar.Header class=" top-0 z-10 gap-4 bg-sidebar/50 p-4 pb-2 backdrop-blur-lg md:sticky">
-		<a href="#/" onclick={handleMobileSidebarItemClick}>
-			<h1 class="inline-flex items-center gap-1 px-2 text-xl font-semibold">llama.cpp UI</h1>
-		</a>
+<div class="flex h-full flex-col">
+	<ScrollArea class="h-full flex-1">
+		<Sidebar.Header class="gap-4 bg-sidebar/50 p-3 backdrop-blur-lg md:pt-4 md:pb-2">
+			<div class="flex items-center justify-between">
+				<a href="#/" onclick={handleMobileSidebarItemClick}>
+					<h1 class="inline-flex items-center gap-1 px-2 text-xl font-semibold">{APP_NAME}</h1>
+				</a>
 
-		<ChatSidebarActions {handleMobileSidebarItemClick} bind:isSearchModeActive bind:searchQuery />
-	</Sidebar.Header>
+				<Button
+					class="rounded-full md:hidden"
+					variant="ghost"
+					size="icon"
+					onclick={() => sidebar.toggle()}
+				>
+					<X class="h-4 w-4" />
+					<span class="sr-only">Close sidebar</span>
+				</Button>
+			</div>
 
-	<Sidebar.Group class="mt-2 space-y-2 p-0 px-4">
-		{#if (filteredConversations.length > 0 && isSearchModeActive) || !isSearchModeActive}
-			<Sidebar.GroupLabel>
-				{isSearchModeActive ? 'Search results' : 'Conversations'}
-			</Sidebar.GroupLabel>
-		{/if}
+			<ChatSidebarActions
+				bind:this={chatSidebarActions}
+				{handleMobileSidebarItemClick}
+				bind:isSearchModeActive
+				bind:searchQuery
+				onSearchDeactivated={handleSearchDeactivated}
+			/>
+		</Sidebar.Header>
 
-		<Sidebar.GroupContent>
-			<Sidebar.Menu>
-				{#each conversationTree as { conversation, depth } (conversation.id)}
-					<Sidebar.MenuItem class="mb-1 p-0">
-						<ChatSidebarConversationItem
-							conversation={{
-								id: conversation.id,
-								name: conversation.name,
-								lastModified: conversation.lastModified,
-								currNode: conversation.currNode,
-								forkedFromConversationId: conversation.forkedFromConversationId
-							}}
-							{depth}
-							{handleMobileSidebarItemClick}
-							isActive={currentChatId === conversation.id}
-							onSelect={selectConversation}
-							onEdit={handleEditConversation}
-							onDelete={handleDeleteConversation}
-							onStop={handleStopGeneration}
-						/>
-					</Sidebar.MenuItem>
-				{/each}
+		<Sidebar.Group class="mt-2 h-[calc(100vh-21rem)] space-y-2 p-0 px-3">
+			{#if (filteredConversations.length > 0 && isSearchModeActive) || !isSearchModeActive}
+				<Sidebar.GroupLabel>
+					{isSearchModeActive ? 'Search results' : 'Recent conversations'}
+				</Sidebar.GroupLabel>
+			{/if}
 
-				{#if conversationTree.length === 0}
-					<div class="px-2 py-4 text-center">
-						<p class="mb-4 p-4 text-sm text-muted-foreground">
-							{searchQuery.length > 0
-								? 'No results found'
-								: isSearchModeActive
-									? 'Start typing to see results'
-									: 'No conversations yet'}
-						</p>
-					</div>
-				{/if}
-			</Sidebar.Menu>
-		</Sidebar.GroupContent>
-	</Sidebar.Group>
-</ScrollArea>
+			<Sidebar.GroupContent>
+				<Sidebar.Menu>
+					{#each conversationTree as { conversation, depth } (conversation.id)}
+						<Sidebar.MenuItem class="mb-1 p-0">
+							<ChatSidebarConversationItem
+								conversation={{
+									id: conversation.id,
+									name: conversation.name,
+									lastModified: conversation.lastModified,
+									currNode: conversation.currNode,
+									forkedFromConversationId: conversation.forkedFromConversationId
+								}}
+								{depth}
+								isActive={currentChatId === conversation.id}
+								onSelect={selectConversation}
+								onEdit={handleEditConversation}
+								onDelete={handleDeleteConversation}
+								onStop={handleStopGeneration}
+							/>
+						</Sidebar.MenuItem>
+					{/each}
+
+					{#if conversationTree.length === 0}
+						<div class="px-2 py-4 text-center">
+							<p class="mb-4 p-4 text-sm text-muted-foreground">
+								{searchQuery.length > 0
+									? 'No results found'
+									: isSearchModeActive
+										? 'Start typing to see results'
+										: 'No conversations yet'}
+							</p>
+						</div>
+					{/if}
+				</Sidebar.Menu>
+			</Sidebar.GroupContent>
+		</Sidebar.Group>
+	</ScrollArea>
+</div>
 
 <DialogConfirmation
 	bind:open={showDeleteDialog}

tools/server/webui/src/lib/components/app/chat/ChatSidebar/ChatSidebarActions.svelte

@@ -1,102 +1,96 @@
 <script lang="ts">
-	import { Search, SquarePen, X } from '@lucide/svelte';
 	import { KeyboardShortcutInfo } from '$lib/components/app';
 	import { Button } from '$lib/components/ui/button';
-	import { Input } from '$lib/components/ui/input';
-	import { McpLogo } from '$lib/components/app';
-	import { SETTINGS_SECTION_TITLES } from '$lib/constants';
-	import { getChatSettingsDialogContext } from '$lib/contexts';
+	import type { Component } from 'svelte';
+	import { SearchInput } from '$lib/components/app';
+	import { page } from '$app/state';
+	import { SIDEBAR_ACTIONS_ITEMS } from '$lib/constants/ui';
 
 	interface Props {
 		handleMobileSidebarItemClick: () => void;
 		isSearchModeActive: boolean;
 		searchQuery: string;
+		isCancelAlwaysVisible?: boolean;
+		onSearchDeactivated?: () => void;
 	}
 
 	let {
 		handleMobileSidebarItemClick,
 		isSearchModeActive = $bindable(),
-		searchQuery = $bindable()
+		searchQuery = $bindable(),
+		isCancelAlwaysVisible = false,
+		onSearchDeactivated
 	}: Props = $props();
 
-	let searchInput: HTMLInputElement | null = $state(null);
-
-	const chatSettingsDialog = getChatSettingsDialogContext();
+	let searchInputRef = $state<HTMLInputElement | null>(null);
 
 	function handleSearchModeDeactivate() {
 		isSearchModeActive = false;
 		searchQuery = '';
+		onSearchDeactivated?.();
 	}
 
-	$effect(() => {
-		if (isSearchModeActive) {
-			searchInput?.focus();
-		}
-	});
+	export function activateSearch() {
+		isSearchModeActive = true;
+		// Focus after Svelte renders the input
+		queueMicrotask(() => searchInputRef?.focus());
+	}
 </script>
 
+{#snippet itemIcon(Icon: Component)}
+	<Icon class="h-4 w-4" />
+{/snippet}
+
 <div class="my-1 space-y-1">
 	{#if isSearchModeActive}
-		<div class="relative">
-			<Search class="absolute top-2.5 left-2 h-4 w-4 text-muted-foreground" />
-
-			<Input
-				bind:ref={searchInput}
-				bind:value={searchQuery}
-				onkeydown={(e) => e.key === 'Escape' && handleSearchModeDeactivate()}
-				placeholder="Search conversations..."
-				class="pl-8"
-			/>
-
-			<X
-				class="cursor-pointertext-muted-foreground absolute top-2.5 right-2 h-4 w-4"
-				onclick={handleSearchModeDeactivate}
-			/>
-		</div>
+		<SearchInput
+			bind:value={searchQuery}
+			bind:ref={searchInputRef}
+			onClose={handleSearchModeDeactivate}
+			onKeyDown={(e) => e.key === 'Escape' && handleSearchModeDeactivate()}
+			placeholder="Search conversations..."
+			{isCancelAlwaysVisible}
+		/>
 	{:else}
-		<Button
-			class="w-full justify-between backdrop-blur-none! hover:[&>kbd]:opacity-100"
-			href="?new_chat=true#/"
-			onclick={handleMobileSidebarItemClick}
-			variant="ghost"
-		>
-			<div class="flex items-center gap-2">
-				<SquarePen class="h-4 w-4" />
+		{#each SIDEBAR_ACTIONS_ITEMS as item (item.route)}
+			{#if !item.route}
+				<Button
+					class="w-full justify-between px-2 backdrop-blur-none! hover:[&>kbd]:opacity-100"
+					onclick={activateSearch}
+					variant="ghost"
+				>
+					<div class="flex items-center gap-2">
+						{@render itemIcon(item.icon)}
 
-				New chat
-			</div>
+						{item.tooltip}
+					</div>
 
-			<KeyboardShortcutInfo keys={['shift', 'cmd', 'o']} />
-		</Button>
+					{#if item.keys}
+						<KeyboardShortcutInfo keys={item.keys} />
+					{/if}
+				</Button>
+			{:else}
+				<Button
+					class="w-full justify-between px-2 backdrop-blur-none! hover:[&>kbd]:opacity-100 {(item.activeRouteId &&
+						page.route.id === item.activeRouteId) ||
+					(item.activeRoutePrefix && page.route.id?.startsWith(item.activeRoutePrefix))
+						? 'bg-accent text-accent-foreground'
+						: ''}"
+					href={item.route}
+					onclick={handleMobileSidebarItemClick}
+					variant="ghost"
+				>
+					<div class="flex items-center gap-2">
+						{@render itemIcon(item.icon)}
 
-		<Button
-			class="w-full justify-between backdrop-blur-none! hover:[&>kbd]:opacity-100"
-			onclick={() => {
-				isSearchModeActive = true;
-			}}
-			variant="ghost"
-		>
-			<div class="flex items-center gap-2">
-				<Search class="h-4 w-4" />
+						{item.tooltip}
+					</div>
 
-				Search
-			</div>
-
-			<KeyboardShortcutInfo keys={['cmd', 'k']} />
-		</Button>
-
-		<Button
-			class="w-full justify-between backdrop-blur-none! hover:[&>kbd]:opacity-100"
-			onclick={() => {
-				chatSettingsDialog.open(SETTINGS_SECTION_TITLES.MCP);
-			}}
-			variant="ghost"
-		>
-			<div class="flex items-center gap-2">
-				<McpLogo class="h-4 w-4" />
-
-				MCP Servers
-			</div>
-		</Button>
+					{#if item.keys}
+						<KeyboardShortcutInfo keys={item.keys} />
+					{/if}
+				</Button>
+			{/if}
+		{/each}
 	{/if}
 </div>

tools/server/webui/src/lib/components/app/chat/ChatSidebar/ChatSidebarConversationItem.svelte

@@ -19,7 +19,6 @@
 		isActive?: boolean;
 		depth?: number;
 		conversation: DatabaseConversation;
-		handleMobileSidebarItemClick?: () => void;
 		onDelete?: (id: string) => void;
 		onEdit?: (id: string) => void;
 		onSelect?: (id: string) => void;
@@ -28,7 +27,6 @@
 
 	let {
 		conversation,
-		handleMobileSidebarItemClick,
 		onDelete,
 		onEdit,
 		onSelect,
@@ -150,9 +148,7 @@
 			</Tooltip.Root>
 		{/if}
 
-		<!-- svelte-ignore a11y_click_events_have_key_events -->
-		<!-- svelte-ignore a11y_no_static_element_interactions -->
-		<span class="truncate text-sm font-medium" onclick={handleMobileSidebarItemClick}>
+		<span class="truncate text-sm font-medium">
 			{conversation.name}
 		</span>
 	</div>

tools/server/webui/src/lib/components/app/chat/index.ts

@@ -124,7 +124,7 @@ export { default as ChatAttachmentsViewAll } from './ChatAttachments/ChatAttachm
  * **Architecture:**
  * - Composes ChatFormTextarea, ChatFormActions, and ChatFormPromptPicker
  * - Manages file upload state via `uploadedFiles` bindable prop
- * - Integrates with ModelsSelector for model selection in router mode
+ * - Integrates with ModelsSelectorDropdown for model selection in router mode
  * - Communicates with parent via callbacks (onSubmit, onFilesAdd, onStop, etc.)
  *
  * **Input Handling:**
@@ -168,14 +168,14 @@ export { default as ChatForm } from './ChatForm/ChatForm.svelte';
  * Images, Text Files, and PDF Files. Each option filters the file picker to
  * appropriate types. Images option is disabled when model lacks vision modality.
  */
-export { default as ChatFormActionAttachmentsDropdown } from './ChatForm/ChatFormActions/ChatFormActionAttachmentsDropdown.svelte';
+export { default as ChatFormActionAttachmentsDropdown } from './ChatForm/ChatFormActions/ChatFormActionsAttachments/ChatFormActionAttachmentsDropdown.svelte';
 
 /**
  * Mobile sheet variant of the file attachment selector. Renders a bottom sheet
  * with the same options as ChatFormActionAttachmentsDropdown, optimized for
  * touch interaction on mobile devices.
  */
-export { default as ChatFormActionAttachmentsSheet } from './ChatForm/ChatFormActions/ChatFormActionAttachmentsSheet.svelte';
+export { default as ChatFormActionAttachmentsSheet } from './ChatForm/ChatFormActions/ChatFormActionsAttachments/ChatFormActionAttachmentsSheet.svelte';
 
 /**
  * Audio recording button with real-time recording indicator. Records audio
@@ -198,6 +198,49 @@ export { default as ChatFormActions } from './ChatForm/ChatFormActions/ChatFormA
  */
 export { default as ChatFormActionSubmit } from './ChatForm/ChatFormActions/ChatFormActionSubmit.svelte';
 
+/**
+ * Dropdown submenu for managing tool permissions in the chat form.
+ *
+ * Displays a collapsible list of available tools organized by group (Built-in / JSON Schema).
+ * Each group can be expanded to show individual tools with checkboxes for enabling/disabling.
+ * Provides bulk enable/disable controls per group and shows enabled/total tool counts.
+ * Opens the tools panel on the server when the menu opens.
+ *
+ * Features:
+ * - Grouped tools with collapsible sections
+ * - Group favicon display (MCP server icons)
+ * - Per-group and per-tool toggle checkboxes
+ * - Loading/error states for tool discovery
+ * - Integration with toolsPanel for state management
+ *
+ * @example
+ * ```svelte
+ * <ChatFormActionToolsSubmenu />
+ * ```
+ */
+export { default as ChatFormActionToolsSubmenu } from './ChatForm/ChatFormActions/ChatFormActionToolsSubmenu.svelte';
+
+/**
+ * Dropdown submenu for managing MCP servers in the chat form.
+ *
+ * Displays a searchable list of enabled MCP servers with toggle switches
+ * to enable/disable each server for chat. Shows server favicon, health status,
+ * and a "Manage MCP Servers" settings link.
+ *
+ * Features:
+ * - Search/filter servers by name or URL
+ * - Per-server toggle to enable/disable for chat
+ * - Health check indicator (shows "Error" badge for failed servers)
+ * - Server favicon display
+ * - Settings link to manage MCP server configuration
+ *
+ * @example
+ * ```svelte
+ * <ChatFormActionMcpServersSubmenu onMcpSettingsClick={handleMcpSettingsClick} />
+ * ```
+ */
+export { default as ChatFormActionMcpServersSubmenu } from './ChatForm/ChatFormActions/ChatFormActionMcpServersSubmenu.svelte';
+
 /**
  * Hidden file input element for programmatic file selection.
  */
@@ -456,6 +499,8 @@ export { default as ChatMessage } from './ChatMessages/ChatMessage.svelte';
  * ```
  */
 export { default as ChatMessageAgenticContent } from './ChatMessages/ChatMessageAgenticContent.svelte';
+export { default as ChatMessagePermissionRequest } from './ChatMessages/ChatMessagePermissionRequest.svelte';
+export { default as ChatMessageContinueRequest } from './ChatMessages/ChatMessageContinueRequest.svelte';
 
 /**
  * Action buttons toolbar for messages. Displays copy, edit, delete, and regenerate
@@ -547,7 +592,7 @@ export { default as ChatMessageEditForm } from './ChatMessages/ChatMessageEditFo
  * and server state. Used as the main content area in chat routes.
  *
  * **Architecture:**
- * - Composes ChatMessages, ChatScreenForm, ChatScreenHeader, and dialogs
+ * - Composes ChatMessages, ChatScreenForm, and dialogs
  * - Manages auto-scroll via `createAutoScrollController()` hook
  * - Handles file upload pipeline (validation → processing → state update)
  * - Integrates with serverStore for loading/error/warning states
@@ -602,13 +647,6 @@ export { default as ChatScreenDragOverlay } from './ChatScreen/ChatScreenDragOve
  */
 export { default as ChatScreenForm } from './ChatScreen/ChatScreenForm.svelte';
 
-/**
- * Header bar for chat screen. Displays conversation title (or "New Chat"),
- * model selector (in router mode), and action buttons (delete conversation).
- * Sticky positioned at the top of the chat area.
- */
-export { default as ChatScreenHeader } from './ChatScreen/ChatScreenHeader.svelte';
-
 /**
  * Processing info display during generation. Shows real-time statistics:
  * tokens per second, prompt/completion token counts, and elapsed time.
@@ -636,55 +674,6 @@ export { default as ChatScreenProcessingInfo } from './ChatScreen/ChatScreenProc
  *
  */
 
-/**
- * **ChatSettings** - Application settings panel
- *
- * Comprehensive settings interface with categorized sections. Manages all
- * user preferences and sampling parameters. Integrates with config store
- * for persistence and ParameterSyncService for server synchronization.
- *
- * **Architecture:**
- * - Uses tabbed navigation with category sections
- * - Maintains local form state, commits on save
- * - Tracks user overrides vs server defaults for sampling params
- * - Exposes reset() method for dialog close without save
- *
- * **Categories:**
- * - **General**: API key, system message, show system messages toggle
- * - **Display**: Theme selection, message actions visibility, model info badge
- * - **Sampling**: Temperature, top_p, top_k, min_p, repeat_penalty, etc.
- * - **Penalties**: Frequency penalty, presence penalty, repeat last N
- * - **Import/Export**: Conversation backup and restore
- * - **MCP**: MCP server management (opens DialogChatSettings with MCP tab)
- * - **Developer**: Debug options, disable auto-scroll
- *
- * **Parameter Sync:**
- * - Fetches defaults from server `/props` endpoint
- * - Shows source indicator badge (Custom/Server Props/Default)
- * - Real-time badge updates as user types
- * - Tracks which parameters user has explicitly overridden
- *
- * **Features:**
- * - Mobile-responsive layout with horizontal scrolling tabs
- * - Form validation with error messages
- * - Secure API key storage (masked input)
- * - Import/export conversations as JSON
- * - Reset to defaults option per parameter
- *
- * **Exported API:**
- * - `reset()` - Reset form fields to currently saved values (for cancel action)
- *
- * @example
- * ```svelte
- * <ChatSettings
- *   bind:this={settingsRef}
- *   onSave={() => dialogOpen = false}
- *   onCancel={() => { settingsRef.reset(); dialogOpen = false; }}
- * />
- * ```
- */
-export { default as ChatSettings } from './ChatSettings/ChatSettings.svelte';
-
 /**
  * Footer with save/cancel buttons for settings panel. Positioned at bottom
  * of settings dialog. Save button commits form state to config store,
@@ -699,13 +688,6 @@ export { default as ChatSettingsFooter } from './ChatSettings/ChatSettingsFooter
  */
 export { default as ChatSettingsFields } from './ChatSettings/ChatSettingsFields.svelte';
 
-/**
- * Import/export tab content for conversation data management. Provides buttons
- * to export all conversations as JSON file and import from JSON file.
- * Handles file download/upload and data validation.
- */
-export { default as ChatSettingsImportExportTab } from './ChatSettings/ChatSettingsImportExportTab.svelte';
-
 /**
  * Badge indicating parameter source for sampling settings. Shows one of:
  * - **Custom**: User has explicitly set this value (orange badge)
@@ -715,6 +697,15 @@ export { default as ChatSettingsImportExportTab } from './ChatSettings/ChatSetti
  */
 export { default as ChatSettingsParameterSourceIndicator } from './ChatSettings/ChatSettingsParameterSourceIndicator.svelte';
 
+/**
+ * **ChatSettingsToolsTab** - Tools configuration tab for chat settings
+ *
+ * Displays available tools grouped by source (built-in, MCP, custom) with
+ * toggles to enable/disable individual tools and tool groups. Shows MCP
+ * server favicons and permission management controls.
+ */
+export { default as ChatSettingsToolsTab } from './ChatSettings/ChatSettingsToolsTab.svelte';
+
 /**
  *
  * SIDEBAR

tools/server/webui/src/lib/components/app/dialogs/DialogChatSettings.svelte

@@ -1,38 +0,0 @@
-<script lang="ts">
-	import * as Dialog from '$lib/components/ui/dialog';
-	import { ChatSettings } from '$lib/components/app';
-	import type { SettingsSectionTitle } from '$lib/constants';
-
-	interface Props {
-		onOpenChange?: (open: boolean) => void;
-		open?: boolean;
-		initialSection?: SettingsSectionTitle;
-	}
-
-	let { onOpenChange, open = false, initialSection }: Props = $props();
-
-	let chatSettingsRef: ChatSettings | undefined = $state();
-
-	function handleClose() {
-		onOpenChange?.(false);
-	}
-
-	function handleSave() {
-		onOpenChange?.(false);
-	}
-
-	$effect(() => {
-		if (open && chatSettingsRef) {
-			chatSettingsRef.reset();
-		}
-	});
-</script>
-
-<Dialog.Root {open} onOpenChange={handleClose}>
-	<Dialog.Content
-		class="z-999999 flex h-[100dvh] max-h-[100dvh] min-h-[100dvh] max-w-4xl! flex-col gap-0 rounded-none
-			p-0 md:h-[64vh] md:max-h-[64vh] md:min-h-0 md:rounded-lg"
-	>
-		<ChatSettings bind:this={chatSettingsRef} onSave={handleSave} {initialSection} />
-	</Dialog.Content>
-</Dialog.Root>

tools/server/webui/src/lib/components/app/dialogs/DialogMcpServerAddNew.svelte

@@ -0,0 +1,88 @@
+<script lang="ts">
+	import { Button } from '$lib/components/ui/button';
+	import * as Dialog from '$lib/components/ui/dialog';
+	import { McpServerForm } from '$lib/components/app/mcp';
+	import { mcpStore } from '$lib/stores/mcp.svelte';
+	import { conversationsStore } from '$lib/stores/conversations.svelte';
+	import { uuid } from '$lib/utils';
+	import { MCP_SERVER_ID_PREFIX } from '$lib/constants';
+
+	interface Props {
+		open: boolean;
+		onOpenChange?: (open: boolean) => void;
+	}
+
+	let { open = $bindable(), onOpenChange }: Props = $props();
+
+	let newServerUrl = $state('');
+	let newServerHeaders = $state('');
+	let newServerUrlError = $derived.by(() => {
+		if (!newServerUrl.trim()) return 'URL is required';
+		try {
+			new URL(newServerUrl);
+
+			return null;
+		} catch {
+			return 'Invalid URL format';
+		}
+	});
+
+	function handleOpenChange(value: boolean) {
+		if (!value) {
+			newServerUrl = '';
+			newServerHeaders = '';
+		}
+		open = value;
+		onOpenChange?.(value);
+	}
+
+	function saveNewServer() {
+		if (newServerUrlError) return;
+
+		const newServerId = uuid() ?? `${MCP_SERVER_ID_PREFIX}-${Date.now()}`;
+
+		mcpStore.addServer({
+			id: newServerId,
+			enabled: true,
+			url: newServerUrl.trim(),
+			headers: newServerHeaders.trim() || undefined
+		});
+
+		conversationsStore.setMcpServerOverride(newServerId, true);
+
+		handleOpenChange(false);
+	}
+</script>
+
+<Dialog.Root {open} onOpenChange={handleOpenChange}>
+	<Dialog.Content class="sm:max-w-md">
+		<Dialog.Header>
+			<Dialog.Title>Add New Server</Dialog.Title>
+		</Dialog.Header>
+
+		<div class="space-y-4 py-4">
+			<McpServerForm
+				url={newServerUrl}
+				headers={newServerHeaders}
+				onUrlChange={(v) => (newServerUrl = v)}
+				onHeadersChange={(v) => (newServerHeaders = v)}
+				urlError={newServerUrl ? newServerUrlError : null}
+				id="new-server"
+			/>
+		</div>
+
+		<Dialog.Footer>
+			<Button variant="secondary" size="sm" onclick={() => handleOpenChange(false)}>Cancel</Button>
+
+			<Button
+				variant="default"
+				size="sm"
+				onclick={saveNewServer}
+				disabled={!!newServerUrlError}
+				aria-label="Save"
+			>
+				Add
+			</Button>
+		</Dialog.Footer>
+	</Dialog.Content>
+</Dialog.Root>

tools/server/webui/src/lib/components/app/dialogs/DialogMcpServersSettings.svelte

@@ -1,39 +0,0 @@
-<script lang="ts">
-	import * as Dialog from '$lib/components/ui/dialog';
-	import { McpLogo, McpServersSettings } from '$lib/components/app';
-
-	interface Props {
-		onOpenChange?: (open: boolean) => void;
-		open?: boolean;
-	}
-
-	let { onOpenChange, open = $bindable(false) }: Props = $props();
-
-	function handleClose() {
-		onOpenChange?.(false);
-	}
-</script>
-
-<Dialog.Root {open} onOpenChange={handleClose}>
-	<Dialog.Content
-		class="z-999999 flex h-[100dvh] max-h-[100dvh] min-h-[100dvh] flex-col gap-0 rounded-none p-0
-			md:h-[80dvh] md:h-auto md:max-h-[80dvh] md:min-h-0 md:rounded-lg"
-		style="max-width: 56rem;"
-	>
-		<div class="grid gap-2 border-b border-border/30 p-4 md:p-6">
-			<Dialog.Title class="inline-flex items-center text-lg font-semibold">
-				<McpLogo class="mr-2 inline h-4 w-4" />
-
-				MCP Servers
-			</Dialog.Title>
-
-			<Dialog.Description class="text-sm text-muted-foreground">
-				Add and configure MCP servers to enable agentic tool execution capabilities.
-			</Dialog.Description>
-		</div>
-
-		<div class="flex-1 overflow-y-auto px-4 py-6">
-			<McpServersSettings />
-		</div>
-	</Dialog.Content>
-</Dialog.Root>

tools/server/webui/src/lib/components/app/dialogs/index.ts

@@ -11,30 +11,12 @@
  */
 
 /**
+ * **DialogMcpServerAddNew** - Add new MCP server dialog
  *
- * SETTINGS DIALOGS
- *
- * Dialogs for application and server configuration.
- *
+ * Modal dialog for adding a new MCP server with URL and optional headers.
+ * Validates URL format and integrates with mcpStore and conversationsStore.
  */
-
-/**
- * **DialogChatSettings** - Settings dialog wrapper
- *
- * Modal dialog containing ChatSettings component with proper
- * open/close state management and automatic form reset on open.
- *
- * **Architecture:**
- * - Wraps ChatSettings component in ShadCN Dialog
- * - Manages open/close state via bindable `open` prop
- * - Resets form state when dialog opens to discard unsaved changes
- *
- * @example
- * ```svelte
- * <DialogChatSettings bind:open={showSettings} />
- * ```
- */
-export { default as DialogChatSettings } from './DialogChatSettings.svelte';
+export { default as DialogMcpServerAddNew } from './DialogMcpServerAddNew.svelte';
 
 /**
  *

tools/server/webui/src/lib/components/app/forms/SearchInput.svelte

@@ -11,6 +11,7 @@
 		class?: string;
 		id?: string;
 		ref?: HTMLInputElement | null;
+		isCancelAlwaysVisible?: boolean;
 	}
 
 	let {
@@ -21,10 +22,11 @@
 		onKeyDown,
 		class: className,
 		id,
-		ref = $bindable(null)
+		ref = $bindable(null),
+		isCancelAlwaysVisible = false
 	}: Props = $props();
 
-	let showClearButton = $derived(!!value || !!onClose);
+	let showClearButton = $derived(isCancelAlwaysVisible || !!value || !!onClose);
 
 	function handleInput(event: Event) {
 		const target = event.target as HTMLInputElement;
@@ -63,7 +65,7 @@
 	{#if showClearButton}
 		<button
 			type="button"
-			class="absolute top-1/2 right-3 -translate-y-1/2 transform text-muted-foreground transition-colors hover:text-foreground"
+			class="absolute top-1/2 right-3 -translate-y-1/2 transform cursor-pointer text-muted-foreground transition-colors hover:text-foreground"
 			onclick={handleClear}
 			aria-label={value ? 'Clear search' : 'Close'}
 		>

tools/server/webui/src/lib/components/app/index.ts

@@ -6,6 +6,7 @@ export * from './dialogs';
 export * from './forms';
 export * from './mcp';
 export * from './misc';
+export * from './settings';
 export * from './models';
 export * from './navigation';
 export * from './server';

tools/server/webui/src/lib/components/app/mcp/McpActiveServersAvatars.svelte

@@ -1,15 +1,18 @@
 <script lang="ts">
 	import { cn } from '$lib/components/ui/utils';
+	import * as Tooltip from '$lib/components/ui/tooltip';
 	import { conversationsStore } from '$lib/stores/conversations.svelte';
 	import { mcpStore } from '$lib/stores/mcp.svelte';
 	import { HealthCheckStatus } from '$lib/enums';
 	import { MAX_DISPLAYED_MCP_AVATARS } from '$lib/constants';
+	import McpLogo from './McpLogo.svelte';
 
 	interface Props {
 		class?: string;
+		onClick?: () => void;
 	}
 
-	let { class: className = '' }: Props = $props();
+	let { class: className = '', onClick }: Props = $props();
 
 	let mcpServers = $derived(mcpStore.getServersSorted().filter((s) => s.enabled));
 	let enabledMcpServersForChat = $derived(
@@ -28,30 +31,60 @@
 	let mcpFavicons = $derived(
 		healthyEnabledMcpServers
 			.slice(0, MAX_DISPLAYED_MCP_AVATARS)
-			.map((s) => ({ id: s.id, url: mcpStore.getServerFavicon(s.id) }))
+			.map((s) => ({
+				id: s.id,
+				name: mcpStore.getServerDisplayName(s.id),
+				url: mcpStore.getServerFavicon(s.id)
+			}))
 			.filter((f) => f.url !== null)
 	);
 </script>
 
-{#if hasEnabledMcpServers && mcpFavicons.length > 0}
-	<div class={cn('inline-flex items-center gap-1.5', className)}>
+{#if !hasEnabledMcpServers}
+	<button
+		class={cn(
+			'inline-flex cursor-pointer items-center gap-0.75 opacity-70 transition-opacity hover:opacity-100',
+			className,
+			'opacity-50 hover:opacity-100'
+		)}
+		onclick={onClick}
+	>
+		<Tooltip.Root>
+			<Tooltip.Trigger>
+				<McpLogo class="h-4 w-4" />
+			</Tooltip.Trigger>
+
+			<Tooltip.Content>
+				<p>MCP Servers</p>
+			</Tooltip.Content>
+		</Tooltip.Root>
+	</button>
+{:else if mcpFavicons.length > 0}
+	<button class={cn('inline-flex items-center gap-0.75', className)} onclick={onClick}>
 		<div class="flex -space-x-1">
 			{#each mcpFavicons as favicon (favicon.id)}
-				<div class="box-shadow-lg overflow-hidden rounded-full bg-muted ring-1 ring-muted">
-					<img
-						src={favicon.url}
-						alt=""
-						class="h-4 w-4"
-						onerror={(e) => {
-							(e.currentTarget as HTMLImageElement).style.display = 'none';
-						}}
-					/>
-				</div>
+				<Tooltip.Root>
+					<Tooltip.Trigger>
+						<div class="box-shadow-lg overflow-hidden rounded-full bg-muted ring-1 ring-muted">
+							<img
+								src={favicon.url}
+								alt=""
+								class="h-4 w-4"
+								onerror={(e) => {
+									(e.currentTarget as HTMLImageElement).style.display = 'none';
+								}}
+							/>
+						</div>
+					</Tooltip.Trigger>
+					<Tooltip.Content>
+						<p>{favicon.name}</p>
+					</Tooltip.Content>
+				</Tooltip.Root>
 			{/each}
 		</div>
 
 		{#if extraServersCount > 0}
 			<span class="text-xs text-muted-foreground">+{extraServersCount}</span>
 		{/if}
-	</div>
+	</button>
 {/if}

tools/server/webui/src/lib/components/app/mcp/McpServerCard/McpServerCardHeader.svelte

@@ -71,7 +71,7 @@
 			</div>
 
 			{#if capabilities || transportType}
-				<div class="flex flex-wrap items-center gap-1">
+				<div class="flex flex-wrap items-center gap-1.5">
 					{#if transportType}
 						{@const TransportIcon = MCP_TRANSPORT_ICONS[transportType]}
 						<Badge variant="outline" class="h-5 gap-1 px-1.5 text-[10px]">

tools/server/webui/src/lib/components/app/mcp/McpServersSettings.svelte

@@ -1,150 +0,0 @@
-<script lang="ts">
-	import { Plus } from '@lucide/svelte';
-	import { Button } from '$lib/components/ui/button';
-	import * as Card from '$lib/components/ui/card';
-	import { uuid } from '$lib/utils';
-	import { mcpStore } from '$lib/stores/mcp.svelte';
-	import { conversationsStore } from '$lib/stores/conversations.svelte';
-	import { McpServerCard, McpServerCardSkeleton, McpServerForm } from '$lib/components/app/mcp';
-	import { MCP_SERVER_ID_PREFIX } from '$lib/constants';
-	import { HealthCheckStatus } from '$lib/enums';
-
-	let servers = $derived(mcpStore.getServersSorted());
-
-	let initialLoadComplete = $state(false);
-
-	$effect(() => {
-		if (initialLoadComplete) return;
-
-		const allChecked =
-			servers.length > 0 &&
-			servers.every((server) => {
-				const state = mcpStore.getHealthCheckState(server.id);
-
-				return (
-					state.status === HealthCheckStatus.SUCCESS || state.status === HealthCheckStatus.ERROR
-				);
-			});
-
-		if (allChecked) {
-			initialLoadComplete = true;
-		}
-	});
-
-	let isAddingServer = $state(false);
-	let newServerUrl = $state('');
-	let newServerHeaders = $state('');
-	let newServerUrlError = $derived.by(() => {
-		if (!newServerUrl.trim()) return 'URL is required';
-		try {
-			new URL(newServerUrl);
-
-			return null;
-		} catch {
-			return 'Invalid URL format';
-		}
-	});
-
-	function showAddServerForm() {
-		isAddingServer = true;
-		newServerUrl = '';
-		newServerHeaders = '';
-	}
-
-	function cancelAddServer() {
-		isAddingServer = false;
-		newServerUrl = '';
-		newServerHeaders = '';
-	}
-
-	function saveNewServer() {
-		if (newServerUrlError) return;
-
-		const newServerId = uuid() ?? `${MCP_SERVER_ID_PREFIX}-${Date.now()}`;
-
-		mcpStore.addServer({
-			id: newServerId,
-			enabled: true,
-			url: newServerUrl.trim(),
-			headers: newServerHeaders.trim() || undefined
-		});
-
-		conversationsStore.setMcpServerOverride(newServerId, true);
-
-		isAddingServer = false;
-		newServerUrl = '';
-		newServerHeaders = '';
-	}
-</script>
-
-<div class="space-y-5 md:space-y-4">
-	<div class="flex items-start justify-between gap-4">
-		<div>
-			<h4 class="text-base font-semibold">Manage Servers</h4>
-		</div>
-
-		{#if !isAddingServer}
-			<Button variant="outline" size="sm" class="shrink-0" onclick={showAddServerForm}>
-				<Plus class="h-4 w-4" />
-
-				Add New Server
-			</Button>
-		{/if}
-	</div>
-
-	{#if isAddingServer}
-		<Card.Root class="bg-muted/30 p-4">
-			<div class="space-y-4">
-				<p class="font-medium">Add New Server</p>
-
-				<McpServerForm
-					url={newServerUrl}
-					headers={newServerHeaders}
-					onUrlChange={(v) => (newServerUrl = v)}
-					onHeadersChange={(v) => (newServerHeaders = v)}
-					urlError={newServerUrl ? newServerUrlError : null}
-					id="new-server"
-				/>
-
-				<div class="flex items-center justify-end gap-2">
-					<Button variant="secondary" size="sm" onclick={cancelAddServer}>Cancel</Button>
-
-					<Button
-						variant="default"
-						size="sm"
-						onclick={saveNewServer}
-						disabled={!!newServerUrlError}
-						aria-label="Save"
-					>
-						Add
-					</Button>
-				</div>
-			</div>
-		</Card.Root>
-	{/if}
-
-	{#if servers.length === 0 && !isAddingServer}
-		<div class="rounded-md border border-dashed p-4 text-sm text-muted-foreground">
-			No MCP Servers configured yet. Add one to enable agentic features.
-		</div>
-	{/if}
-
-	{#if servers.length > 0}
-		<div class="space-y-3">
-			{#each servers as server (server.id)}
-				{#if !initialLoadComplete}
-					<McpServerCardSkeleton />
-				{:else}
-					<McpServerCard
-						{server}
-						faviconUrl={mcpStore.getServerFavicon(server.id)}
-						enabled={conversationsStore.isMcpServerEnabledForChat(server.id)}
-						onToggle={async () => await conversationsStore.toggleMcpServerForChat(server.id)}
-						onUpdate={(updates) => mcpStore.updateServer(server.id, updates)}
-						onDelete={() => mcpStore.removeServer(server.id)}
-					/>
-				{/if}
-			{/each}
-		</div>
-	{/if}
-</div>

tools/server/webui/src/lib/components/app/mcp/index.ts

@@ -39,7 +39,7 @@
  * <McpServersSettings />
  * ```
  */
-export { default as McpServersSettings } from './McpServersSettings.svelte';
+export { default as McpServersSettings } from '../settings/SettingsMcpServers.svelte';
 
 /**
  * **McpActiveServersAvatars** - Active MCP servers indicator
@@ -69,33 +69,6 @@ export { default as McpServersSettings } from './McpServersSettings.svelte';
  */
 export { default as McpActiveServersAvatars } from './McpActiveServersAvatars.svelte';
 
-/**
- * **McpServersSelector** - Quick MCP server toggle dropdown
- *
- * Compact dropdown for quickly enabling/disabling MCP servers for the current chat.
- * Uses McpActiveServersAvatars as trigger and shows searchable server list with switches.
- *
- * **Architecture:**
- * - Uses DropdownMenuSearchable for searchable dropdown UI
- * - McpActiveServersAvatars as the trigger element
- * - Integrates with conversationsStore for per-chat toggle
- * - Runs health checks on dropdown open
- *
- * **Features:**
- * - Searchable server list by name/URL
- * - Switch toggles matching McpServersSettings behavior
- * - Error state display for unhealthy servers
- * - Footer link to full MCP settings dialog
- *
- * @example
- * ```svelte
- * <McpServersSelector
- *   onSettingsClick={() => showMcpSettings = true}
- * />
- * ```
- */
-export { default as McpServersSelector } from './McpServersSelector.svelte';
-
 /**
  * **McpCapabilitiesBadges** - Server capabilities display
  *

tools/server/webui/src/lib/components/app/misc/ConversationSelection.svelte

@@ -1,8 +1,7 @@
 <script lang="ts">
-	import { Search, X } from '@lucide/svelte';
 	import { Button } from '$lib/components/ui/button';
-	import { Input } from '$lib/components/ui/input';
 	import { Checkbox } from '$lib/components/ui/checkbox';
+	import SearchInput from '$lib/components/app/forms/SearchInput.svelte';
 	import { ScrollArea } from '$lib/components/ui/scroll-area';
 	import { SvelteSet } from 'svelte/reactivity';
 
@@ -111,21 +110,7 @@
 </script>
 
 <div class="space-y-4">
-	<div class="relative">
-		<Search class="absolute top-1/2 left-3 h-4 w-4 -translate-y-1/2 text-muted-foreground" />
-
-		<Input bind:value={searchQuery} placeholder="Search conversations..." class="pr-9 pl-9" />
-
-		{#if searchQuery}
-			<button
-				class="absolute top-1/2 right-3 -translate-y-1/2 text-muted-foreground hover:text-foreground"
-				onclick={() => (searchQuery = '')}
-				type="button"
-			>
-				<X class="h-4 w-4" />
-			</button>
-		{/if}
-	</div>
+	<SearchInput bind:value={searchQuery} placeholder="Search conversations..." />
 
 	<div class="flex items-center justify-between text-sm text-muted-foreground">
 		<span>

tools/server/webui/src/lib/components/app/models/ModelId.svelte

@@ -5,8 +5,9 @@
 
 	interface Props {
 		modelId: string;
-		showOrgName?: boolean;
+		hideOrgName?: boolean;
 		showRaw?: boolean;
+		hideQuantization?: boolean;
 		aliases?: string[];
 		tags?: string[];
 		class?: string;
@@ -14,8 +15,9 @@
 
 	let {
 		modelId,
-		showOrgName = false,
+		hideOrgName = false,
 		showRaw = undefined,
+		hideQuantization = false,
 		aliases,
 		tags,
 		class: className = '',
@@ -41,7 +43,7 @@
 {:else}
 	<span class="flex min-w-0 flex-wrap items-center gap-1 {className}" {...rest}>
 		<span class="min-w-0 truncate font-medium">
-			{#if showOrgName && parsed.orgName && !(aliases && aliases.length > 0)}{parsed.orgName}/{/if}{displayName}
+			{#if !hideOrgName && parsed.orgName && !(aliases && aliases.length > 0)}{parsed.orgName}/{/if}{displayName}
 		</span>
 
 		{#if parsed.params}
@@ -50,7 +52,7 @@
 			</span>
 		{/if}
 
-		{#if parsed.quantization}
+		{#if parsed.quantization && !hideQuantization}
 			<span class={badgeClass}>
 				{parsed.quantization}
 			</span>

tools/server/webui/src/lib/components/app/models/ModelsSelectorDropdown.svelte

@@ -1,19 +1,10 @@
 <script lang="ts">
-	import { onMount } from 'svelte';
 	import { ChevronDown, Loader2, Package } from '@lucide/svelte';
 	import * as DropdownMenu from '$lib/components/ui/dropdown-menu';
 	import * as Tooltip from '$lib/components/ui/tooltip';
 	import { cn } from '$lib/components/ui/utils';
-	import {
-		modelsStore,
-		modelOptions,
-		modelsLoading,
-		modelsUpdating,
-		selectedModelId,
-		singleModelName
-	} from '$lib/stores/models.svelte';
 	import { KeyboardKey } from '$lib/enums';
-	import { isRouterMode } from '$lib/stores/server.svelte';
+	import { useModelsSelector } from '$lib/hooks/use-models-selector.svelte';
 	import {
 		DialogModelInformation,
 		DropdownMenuSearchable,
@@ -21,8 +12,7 @@
 		ModelsSelectorList,
 		ModelsSelectorOption
 	} from '$lib/components/app';
-	import type { ModelOption } from '$lib/types/models';
-	import { filterModelOptions, groupModelOptions, type ModelItem } from './utils';
+	import type { ModelItem } from './utils';
 
 	interface Props {
 		class?: string;
@@ -42,90 +32,26 @@
 		useGlobalSelection = false
 	}: Props = $props();
 
-	let options = $derived(
-		modelOptions().filter((option) => {
-			const modelProps = modelsStore.getModelProps(option.model);
-
-			return modelProps?.webui !== false;
-		})
-	);
-	let loading = $derived(modelsLoading());
-	let updating = $derived(modelsUpdating());
-	let activeId = $derived(selectedModelId());
-	let isRouter = $derived(isRouterMode());
-	let serverModel = $derived(singleModelName());
-
-	let isHighlightedCurrentModelActive = $derived.by(() => {
-		if (!isRouter || !currentModel) return false;
-
-		const currentOption = options.find((option) => option.model === currentModel);
-
-		return currentOption ? currentOption.id === activeId : false;
-	});
-
-	let isCurrentModelInCache = $derived.by(() => {
-		if (!isRouter || !currentModel) return true;
-
-		return options.some((option) => option.model === currentModel);
-	});
-
-	let isLoadingModel = $state(false);
-
-	let searchTerm = $state('');
+	let isOpen = $state(false);
 	let highlightedIndex = $state<number>(-1);
 
-	let filteredOptions = $derived(filterModelOptions(options, searchTerm));
-
-	let groupedFilteredOptions = $derived(
-		groupModelOptions(filteredOptions, modelsStore.favoriteModelIds, (m) =>
-			modelsStore.isModelLoaded(m)
-		)
-	);
+	const ms = useModelsSelector({
+		currentModel: () => currentModel,
+		useGlobalSelection: () => useGlobalSelection,
+		onModelChange: () => onModelChange,
+		onOpenChange: (open) => {
+			isOpen = open;
+			highlightedIndex = -1;
+		}
+	});
 
 	$effect(() => {
-		void searchTerm;
+		void ms.searchTerm;
 		highlightedIndex = -1;
 	});
 
-	let isOpen = $state(false);
-	let showModelDialog = $state(false);
-	let infoModelId = $state<string | null>(null);
-
-	function handleInfoClick(modelName: string) {
-		infoModelId = modelName;
-		showModelDialog = true;
-	}
-
-	onMount(() => {
-		modelsStore.fetch().catch((error) => {
-			console.error('Unable to load models:', error);
-		});
-	});
-
-	function handleOpenChange(open: boolean) {
-		if (loading || updating) return;
-
-		if (isRouter) {
-			if (open) {
-				isOpen = true;
-				searchTerm = '';
-				highlightedIndex = -1;
-
-				modelsStore.fetchRouterModels().then(() => {
-					modelsStore.fetchModalitiesForLoadedModels();
-				});
-			} else {
-				isOpen = false;
-				searchTerm = '';
-				highlightedIndex = -1;
-			}
-		} else {
-			showModelDialog = open;
-		}
-	}
-
 	export function open() {
-		handleOpenChange(true);
+		ms.handleOpenChange(true);
 	}
 
 	function handleSearchKeyDown(event: KeyboardEvent) {
@@ -134,9 +60,9 @@
 		if (event.key === KeyboardKey.ARROW_DOWN) {
 			event.preventDefault();
 
-			if (filteredOptions.length === 0) return;
+			if (ms.filteredOptions.length === 0) return;
 
-			if (highlightedIndex === -1 || highlightedIndex === filteredOptions.length - 1) {
+			if (highlightedIndex === -1 || highlightedIndex === ms.filteredOptions.length - 1) {
 				highlightedIndex = 0;
 			} else {
 				highlightedIndex += 1;
@@ -144,146 +70,69 @@
 		} else if (event.key === KeyboardKey.ARROW_UP) {
 			event.preventDefault();
 
-			if (filteredOptions.length === 0) return;
+			if (ms.filteredOptions.length === 0) return;
 
 			if (highlightedIndex === -1 || highlightedIndex === 0) {
-				highlightedIndex = filteredOptions.length - 1;
+				highlightedIndex = ms.filteredOptions.length - 1;
 			} else {
 				highlightedIndex -= 1;
 			}
 		} else if (event.key === KeyboardKey.ENTER) {
 			event.preventDefault();
 
-			if (highlightedIndex >= 0 && highlightedIndex < filteredOptions.length) {
-				const option = filteredOptions[highlightedIndex];
+			if (highlightedIndex >= 0 && highlightedIndex < ms.filteredOptions.length) {
+				const option = ms.filteredOptions[highlightedIndex];
 
-				handleSelect(option.id);
-			} else if (filteredOptions.length > 0) {
+				ms.handleSelect(option.id);
+			} else if (ms.filteredOptions.length > 0) {
 				highlightedIndex = 0;
 			}
 		}
 	}
-
-	async function handleSelect(modelId: string) {
-		const option = options.find((opt) => opt.id === modelId);
-		if (!option) return;
-
-		let shouldCloseMenu = true;
-
-		if (onModelChange) {
-			const result = await onModelChange(option.id, option.model);
-
-			if (result === false) {
-				shouldCloseMenu = false;
-			}
-		} else {
-			await modelsStore.selectModelById(option.id);
-		}
-
-		if (shouldCloseMenu) {
-			handleOpenChange(false);
-
-			requestAnimationFrame(() => {
-				const textarea = document.querySelector<HTMLTextAreaElement>(
-					'[data-slot="chat-form"] textarea'
-				);
-				textarea?.focus();
-			});
-		}
-
-		if (!onModelChange && isRouter && !modelsStore.isModelLoaded(option.model)) {
-			isLoadingModel = true;
-			modelsStore
-				.loadModel(option.model)
-				.catch((error) => console.error('Failed to load model:', error))
-				.finally(() => (isLoadingModel = false));
-		}
-	}
-
-	function getDisplayOption(): ModelOption | undefined {
-		if (!isRouter) {
-			const displayModel = serverModel || currentModel;
-			if (displayModel) {
-				return {
-					id: serverModel ? 'current' : 'offline-current',
-					model: displayModel,
-					name: displayModel.split('/').pop() || displayModel,
-					capabilities: []
-				};
-			}
-
-			return undefined;
-		}
-
-		if (useGlobalSelection && activeId) {
-			const selected = options.find((option) => option.id === activeId);
-
-			if (selected) return selected;
-		}
-
-		if (currentModel) {
-			if (!isCurrentModelInCache) {
-				return {
-					id: 'not-in-cache',
-					model: currentModel,
-					name: currentModel.split('/').pop() || currentModel,
-					capabilities: []
-				};
-			}
-
-			return options.find((option) => option.model === currentModel);
-		}
-
-		if (activeId) {
-			return options.find((option) => option.id === activeId);
-		}
-
-		return undefined;
-	}
 </script>
 
 <div class={cn('relative inline-flex flex-col items-end gap-1', className)}>
-	{#if loading && options.length === 0 && isRouter}
+	{#if ms.loading && ms.options.length === 0 && ms.isRouter}
 		<div class="flex items-center gap-2 text-xs text-muted-foreground">
 			<Loader2 class="h-3.5 w-3.5 animate-spin" />
 
 			Loading models…
 		</div>
-	{:else if options.length === 0 && isRouter}
+	{:else if ms.options.length === 0 && ms.isRouter}
 		{#if currentModel}
 			<span
 				class={cn(
 					'inline-flex items-center gap-1.5 rounded-sm bg-muted-foreground/10 px-1.5 py-1 text-xs text-muted-foreground',
 					className
 				)}
-				style="max-width: min(calc(100cqw - 9rem), 20rem)"
+				style="max-width: min(calc(100cqw - 10rem), 20rem)"
 			>
 				<Package class="h-3.5 w-3.5" />
 
-				<ModelId modelId={currentModel} class="min-w-0" showOrgName />
+				<ModelId modelId={currentModel} class="min-w-0" hideQuantization />
 			</span>
 		{:else}
 			<p class="text-xs text-muted-foreground">No models available.</p>
 		{/if}
 	{:else}
-		{@const selectedOption = getDisplayOption()}
+		{@const selectedOption = ms.getDisplayOption()}
 
-		{#if isRouter}
-			<DropdownMenu.Root bind:open={isOpen} onOpenChange={handleOpenChange}>
+		{#if ms.isRouter}
+			<DropdownMenu.Root bind:open={isOpen} onOpenChange={ms.handleOpenChange}>
 				<DropdownMenu.Trigger
 					class={cn(
 						`inline-grid cursor-pointer grid-cols-[1fr_auto_1fr] items-center gap-1.5 rounded-sm bg-muted-foreground/10 px-1.5 py-1 text-xs transition hover:text-foreground focus:outline-none focus-visible:ring-2 focus-visible:ring-ring focus-visible:ring-offset-2 disabled:cursor-not-allowed disabled:opacity-60`,
-						!isCurrentModelInCache
+						!ms.isCurrentModelInCache
 							? 'bg-red-400/10 !text-red-400 hover:bg-red-400/20 hover:text-red-400'
 							: forceForegroundText
 								? 'text-foreground'
-								: isHighlightedCurrentModelActive
+								: ms.isHighlightedCurrentModelActive
 									? 'text-foreground'
 									: 'text-muted-foreground',
-						isOpen ? 'text-foreground' : ''
+						isOpen ? 'text-foreground' : '',
+						'max-w-[min(calc(100vw-4rem) md:max-w-[min(calc(100cqw-9rem),25rem)]'
 					)}
-					style="max-width: min(calc(100cqw - 9rem), 20rem)"
-					disabled={disabled || updating}
+					disabled={disabled || ms.updating}
 				>
 					<Package class="h-3.5 w-3.5" />
 
@@ -295,7 +144,7 @@
 									<ModelId
 										modelId={selectedOption.model}
 										class="min-w-0 overflow-hidden"
-										showOrgName
+										hideOrgName={false}
 										{...props}
 									/>
 								{/snippet}
@@ -309,7 +158,7 @@
 						<span class="min-w-0 font-medium">Select model</span>
 					{/if}
 
-					{#if updating || isLoadingModel}
+					{#if ms.updating || ms.isLoadingModel}
 						<Loader2 class="h-3 w-3.5 animate-spin" />
 					{:else}
 						<ChevronDown class="h-3 w-3.5" />
@@ -321,14 +170,15 @@
 					class="w-full max-w-[100vw] pt-0 sm:w-max sm:max-w-[calc(100vw-2rem)]"
 				>
 					<DropdownMenuSearchable
-						bind:searchValue={searchTerm}
+						searchValue={ms.searchTerm}
+						onSearchChange={(v) => ms.setSearchTerm(v)}
 						placeholder="Search models..."
 						onSearchKeyDown={handleSearchKeyDown}
 						emptyMessage="No models found."
-						isEmpty={filteredOptions.length === 0 && isCurrentModelInCache}
+						isEmpty={ms.filteredOptions.length === 0 && ms.isCurrentModelInCache}
 					>
 						<div class="models-list">
-							{#if !isCurrentModelInCache && currentModel}
+							{#if !ms.isCurrentModelInCache && currentModel}
 								<!-- Show unavailable model as first option (disabled) -->
 								<button
 									type="button"
@@ -338,47 +188,47 @@
 									aria-disabled="true"
 									disabled
 								>
-									<ModelId modelId={currentModel} class="flex-1" showOrgName />
+									<ModelId modelId={currentModel} class="flex-1" hideQuantization />
 
 									<span class="ml-2 text-xs whitespace-nowrap opacity-70">(not available)</span>
 								</button>
 							{/if}
 
-							{#if filteredOptions.length === 0}
+							{#if ms.filteredOptions.length === 0}
 								<p class="px-4 py-3 text-sm text-muted-foreground">No models found.</p>
 							{/if}
 
-							{#snippet modelOption(item: ModelItem, showOrgName: boolean)}
+							{#snippet modelOption(item: ModelItem, hideOrgName: boolean)}
 								{@const { option, flatIndex } = item}
-								{@const isSelected = currentModel === option.model || activeId === option.id}
+								{@const isSelected = currentModel === option.model || ms.activeId === option.id}
 								{@const isHighlighted = flatIndex === highlightedIndex}
-								{@const isFav = modelsStore.favoriteModelIds.has(option.model)}
+								{@const isFav = ms.isFavorite(option.model)}
 
 								<ModelsSelectorOption
 									{option}
 									{isSelected}
 									{isHighlighted}
 									{isFav}
-									{showOrgName}
-									onSelect={handleSelect}
-									onInfoClick={handleInfoClick}
+									{hideOrgName}
+									onSelect={ms.handleSelect}
+									onInfoClick={ms.handleInfoClick}
 									onMouseEnter={() => (highlightedIndex = flatIndex)}
 									onKeyDown={(e) => {
 										if (e.key === KeyboardKey.ENTER || e.key === KeyboardKey.SPACE) {
 											e.preventDefault();
-											handleSelect(option.id);
+											ms.handleSelect(option.id);
 										}
 									}}
 								/>
 							{/snippet}
 
 							<ModelsSelectorList
-								groups={groupedFilteredOptions}
+								groups={ms.groupedFilteredOptions}
 								{currentModel}
-								{activeId}
+								activeId={ms.activeId}
 								sectionHeaderClass="my-1.5 px-2 py-2 text-[13px] font-semibold text-muted-foreground/70 select-none"
-								onSelect={handleSelect}
-								onInfoClick={handleInfoClick}
+								onSelect={ms.handleSelect}
+								onInfoClick={ms.handleInfoClick}
 								renderOption={modelOption}
 							/>
 						</div>
@@ -389,18 +239,18 @@
 			<button
 				class={cn(
 					`inline-flex cursor-pointer items-center gap-1.5 rounded-sm bg-muted-foreground/10 px-1.5 py-1 text-xs transition hover:text-foreground focus:outline-none focus-visible:ring-2 focus-visible:ring-ring focus-visible:ring-offset-2 disabled:cursor-not-allowed disabled:opacity-60`,
-					!isCurrentModelInCache
+					!ms.isCurrentModelInCache
 						? 'bg-red-400/10 !text-red-400 hover:bg-red-400/20 hover:text-red-400'
 						: forceForegroundText
 							? 'text-foreground'
-							: isHighlightedCurrentModelActive
+							: ms.isHighlightedCurrentModelActive
 								? 'text-foreground'
 								: 'text-muted-foreground',
 					isOpen ? 'text-foreground' : ''
 				)}
 				style="max-width: min(calc(100cqw - 6.5rem), 32rem)"
-				onclick={() => handleOpenChange(true)}
-				disabled={disabled || updating}
+				onclick={() => ms.handleOpenChange(true)}
+				disabled={disabled || ms.updating}
 			>
 				<Package class="h-3.5 w-3.5" />
 
@@ -412,7 +262,7 @@
 								<ModelId
 									modelId={selectedOption.model}
 									class="min-w-0 overflow-hidden"
-									showOrgName
+									hideOrgName={false}
 									{...props}
 								/>
 							{/snippet}
@@ -424,7 +274,7 @@
 					</Tooltip.Root>
 				{/if}
 
-				{#if updating}
+				{#if ms.updating}
 					<Loader2 class="h-3 w-3.5 animate-spin" />
 				{/if}
 			</button>
@@ -432,6 +282,10 @@
 	{/if}
 </div>
 
-{#if showModelDialog}
-	<DialogModelInformation bind:open={showModelDialog} modelId={infoModelId} />
+{#if ms.showModelDialog}
+	<DialogModelInformation
+		open={ms.showModelDialog}
+		onOpenChange={(v) => ms.setShowModelDialog(v)}
+		modelId={ms.infoModelId}
+	/>
 {/if}

tools/server/webui/src/lib/components/app/models/ModelsSelectorList.svelte

@@ -27,7 +27,7 @@
 	let render = $derived(renderOption ?? defaultOption);
 </script>
 
-{#snippet defaultOption(item: ModelItem, showOrgName: boolean)}
+{#snippet defaultOption(item: ModelItem, hideOrgName: boolean)}
 	{@const { option } = item}
 	{@const isSelected = currentModel === option.model || activeId === option.id}
 	{@const isFav = modelsStore.favoriteModelIds.has(option.model)}
@@ -37,7 +37,7 @@
 		{isSelected}
 		isHighlighted={false}
 		{isFav}
-		{showOrgName}
+		{hideOrgName}
 		{onSelect}
 		{onInfoClick}
 		onMouseEnter={() => {}}
@@ -48,7 +48,7 @@
 {#if groups.loaded.length > 0}
 	<p class={sectionHeaderClass}>Loaded models</p>
 	{#each groups.loaded as item (`loaded-${item.option.id}`)}
-		{@render render(item, true)}
+		{@render render(item, false)}
 	{/each}
 {/if}
 
@@ -66,7 +66,7 @@
 			<p class={orgHeaderClass}>{group.orgName}</p>
 		{/if}
 		{#each group.items as item (item.option.id)}
-			{@render render(item, false)}
+			{@render render(item, true)}
 		{/each}
 	{/each}
 {/if}

tools/server/webui/src/lib/components/app/models/ModelsSelectorOption.svelte

@@ -20,7 +20,7 @@
 		isSelected: boolean;
 		isHighlighted: boolean;
 		isFav: boolean;
-		showOrgName?: boolean;
+		hideOrgName?: boolean;
 		onSelect: (modelId: string) => void;
 		onMouseEnter: () => void;
 		onKeyDown: (e: KeyboardEvent) => void;
@@ -32,7 +32,7 @@
 		isSelected,
 		isHighlighted,
 		isFav,
-		showOrgName = false,
+		hideOrgName = false,
 		onSelect,
 		onMouseEnter,
 		onKeyDown,
@@ -71,7 +71,7 @@
 >
 	<ModelId
 		modelId={option.model}
-		{showOrgName}
+		{hideOrgName}
 		aliases={option.aliases}
 		tags={option.tags}
 		class="flex-1"

tools/server/webui/src/lib/components/app/models/ModelsSelectorSheet.svelte

@@ -1,25 +1,15 @@
 <script lang="ts">
-	import { onMount } from 'svelte';
 	import { ChevronDown, Loader2, Package } from '@lucide/svelte';
 	import * as Sheet from '$lib/components/ui/sheet';
 	import { cn } from '$lib/components/ui/utils';
-	import {
-		modelsStore,
-		modelOptions,
-		modelsLoading,
-		modelsUpdating,
-		selectedModelId,
-		singleModelName
-	} from '$lib/stores/models.svelte';
-	import { isRouterMode } from '$lib/stores/server.svelte';
+	import { useModelsSelector } from '$lib/hooks/use-models-selector.svelte';
 	import {
 		DialogModelInformation,
+		ModelId,
 		ModelsSelectorList,
 		SearchInput,
 		TruncatedText
 	} from '$lib/components/app';
-	import type { ModelOption } from '$lib/types/models';
-	import { filterModelOptions, groupModelOptions } from './utils';
 
 	interface Props {
 		class?: string;
@@ -41,201 +31,71 @@
 		useGlobalSelection = false
 	}: Props = $props();
 
-	let options = $derived(
-		modelOptions().filter((option) => {
-			const modelProps = modelsStore.getModelProps(option.model);
-			return modelProps?.webui !== false;
-		})
-	);
-	let loading = $derived(modelsLoading());
-	let updating = $derived(modelsUpdating());
-	let activeId = $derived(selectedModelId());
-	let isRouter = $derived(isRouterMode());
-	let serverModel = $derived(singleModelName());
-
-	let isLoadingModel = $state(false);
-
-	let isHighlightedCurrentModelActive = $derived(
-		!isRouter || !currentModel
-			? false
-			: (() => {
-					const currentOption = options.find((option) => option.model === currentModel);
-
-					return currentOption ? currentOption.id === activeId : false;
-				})()
-	);
-
-	let isCurrentModelInCache = $derived.by(() => {
-		if (!isRouter || !currentModel) return true;
-
-		return options.some((option) => option.model === currentModel);
-	});
-
-	let searchTerm = $state('');
-
-	let filteredOptions = $derived(filterModelOptions(options, searchTerm));
-
-	let groupedFilteredOptions = $derived(
-		groupModelOptions(filteredOptions, modelsStore.favoriteModelIds, (m) =>
-			modelsStore.isModelLoaded(m)
-		)
-	);
-
 	let sheetOpen = $state(false);
-	let showModelDialog = $state(false);
-	let infoModelId = $state<string | null>(null);
 
-	function handleInfoClick(modelName: string) {
-		infoModelId = modelName;
-		showModelDialog = true;
-	}
-
-	onMount(() => {
-		modelsStore.fetch().catch((error) => {
-			console.error('Unable to load models:', error);
-		});
-	});
-
-	function handleOpenChange(open: boolean) {
-		if (loading || updating) return;
-
-		if (isRouter) {
-			if (open) {
-				sheetOpen = true;
-				searchTerm = '';
-
-				modelsStore.fetchRouterModels().then(() => {
-					modelsStore.fetchModalitiesForLoadedModels();
-				});
-			} else {
-				sheetOpen = false;
-				searchTerm = '';
-			}
-		} else {
-			showModelDialog = open;
+	const ms = useModelsSelector({
+		currentModel: () => currentModel,
+		useGlobalSelection: () => useGlobalSelection,
+		onModelChange: () => onModelChange,
+		onOpenChange: (open) => {
+			sheetOpen = open;
 		}
-	}
+	});
 
 	export function open() {
-		handleOpenChange(true);
+		ms.handleOpenChange(true);
 	}
 
 	function handleSheetOpenChange(open: boolean) {
 		if (!open) {
-			handleOpenChange(false);
+			ms.handleOpenChange(false);
 		}
 	}
-
-	async function handleSelect(modelId: string) {
-		const option = options.find((opt) => opt.id === modelId);
-		if (!option) return;
-
-		let shouldCloseMenu = true;
-
-		if (onModelChange) {
-			const result = await onModelChange(option.id, option.model);
-
-			if (result === false) {
-				shouldCloseMenu = false;
-			}
-		} else {
-			await modelsStore.selectModelById(option.id);
-		}
-
-		if (shouldCloseMenu) {
-			handleOpenChange(false);
-
-			requestAnimationFrame(() => {
-				const textarea = document.querySelector<HTMLTextAreaElement>(
-					'[data-slot="chat-form"] textarea'
-				);
-				textarea?.focus();
-			});
-		}
-
-		if (!onModelChange && isRouter && !modelsStore.isModelLoaded(option.model)) {
-			isLoadingModel = true;
-			modelsStore
-				.loadModel(option.model)
-				.catch((error) => console.error('Failed to load model:', error))
-				.finally(() => (isLoadingModel = false));
-		}
-	}
-
-	function getDisplayOption(): ModelOption | undefined {
-		if (!isRouter) {
-			if (serverModel) {
-				return {
-					id: 'current',
-					model: serverModel,
-					name: serverModel.split('/').pop() || serverModel,
-					capabilities: []
-				};
-			}
-
-			return undefined;
-		}
-
-		if (useGlobalSelection && activeId) {
-			const selected = options.find((option) => option.id === activeId);
-			if (selected) return selected;
-		}
-
-		if (currentModel) {
-			if (!isCurrentModelInCache) {
-				return {
-					id: 'not-in-cache',
-					model: currentModel,
-					name: currentModel.split('/').pop() || currentModel,
-					capabilities: []
-				};
-			}
-
-			return options.find((option) => option.model === currentModel);
-		}
-
-		if (activeId) {
-			return options.find((option) => option.id === activeId);
-		}
-
-		return undefined;
-	}
 </script>
 
 <div class={cn('relative inline-flex flex-col items-end gap-1', className)}>
-	{#if loading && options.length === 0 && isRouter}
+	{#if ms.loading && ms.options.length === 0 && ms.isRouter}
 		<div class="flex items-center gap-2 text-xs text-muted-foreground">
 			<Loader2 class="h-3.5 w-3.5 animate-spin" />
 			Loading models…
 		</div>
-	{:else if options.length === 0 && isRouter}
+	{:else if ms.options.length === 0 && ms.isRouter}
 		<p class="text-xs text-muted-foreground">No models available.</p>
 	{:else}
-		{@const selectedOption = getDisplayOption()}
+		{@const selectedOption = ms.getDisplayOption()}
 
-		{#if isRouter}
+		{#if ms.isRouter}
 			<button
 				type="button"
 				class={cn(
 					`inline-grid cursor-pointer grid-cols-[1fr_auto_1fr] items-center gap-1.5 rounded-sm bg-muted-foreground/10 px-1.5 py-1 text-xs transition hover:text-foreground focus:outline-none focus-visible:ring-2 focus-visible:ring-ring focus-visible:ring-offset-2 disabled:cursor-not-allowed disabled:opacity-60`,
-					!isCurrentModelInCache
+					!ms.isCurrentModelInCache
 						? 'bg-red-400/10 !text-red-400 hover:bg-red-400/20 hover:text-red-400'
 						: forceForegroundText
 							? 'text-foreground'
-							: isHighlightedCurrentModelActive
+							: ms.isHighlightedCurrentModelActive
 								? 'text-foreground'
 								: 'text-muted-foreground',
 					sheetOpen ? 'text-foreground' : ''
 				)}
 				style="max-width: min(calc(100cqw - 9rem), 20rem)"
-				disabled={disabled || updating}
-				onclick={() => handleOpenChange(true)}
+				disabled={disabled || ms.updating}
+				onclick={() => ms.handleOpenChange(true)}
 			>
 				<Package class="h-3.5 w-3.5" />
 
-				<TruncatedText text={selectedOption?.model || 'Select model'} class="min-w-0 font-medium" />
+				{#if !selectedOption}
+					<span class="min-w-0 font-medium">Select model</span>
+				{:else}
+					<ModelId
+						class="text-xs"
+						modelId={selectedOption?.model || ''}
+						hideQuantization
+						hideOrgName
+					/>
+				{/if}
 
-				{#if updating || isLoadingModel}
+				{#if ms.updating || ms.isLoadingModel}
 					<Loader2 class="h-3 w-3.5 animate-spin" />
 				{:else}
 					<ChevronDown class="h-3 w-3.5" />
@@ -254,11 +114,15 @@
 
 					<div class="flex flex-col gap-1 pb-4">
 						<div class="mb-3 px-4">
-							<SearchInput placeholder="Search models..." bind:value={searchTerm} />
+							<SearchInput
+								placeholder="Search models..."
+								value={ms.searchTerm}
+								onInput={(v) => ms.setSearchTerm(v)}
+							/>
 						</div>
 
 						<div class="max-h-[60vh] overflow-y-auto px-2">
-							{#if !isCurrentModelInCache && currentModel}
+							{#if !ms.isCurrentModelInCache && currentModel}
 								<button
 									type="button"
 									class="flex w-full cursor-not-allowed items-center rounded-md bg-red-400/10 px-3 py-2.5 text-left text-sm text-red-400"
@@ -272,18 +136,18 @@
 								<div class="my-1 h-px bg-border"></div>
 							{/if}
 
-							{#if filteredOptions.length === 0}
+							{#if ms.filteredOptions.length === 0}
 								<p class="px-3 py-3 text-center text-sm text-muted-foreground">No models found.</p>
 							{/if}
 
 							<ModelsSelectorList
-								groups={groupedFilteredOptions}
+								groups={ms.groupedFilteredOptions}
 								{currentModel}
-								{activeId}
+								activeId={ms.activeId}
 								sectionHeaderClass="px-2 py-2 text-xs font-semibold text-muted-foreground/60 select-none"
 								orgHeaderClass="px-2 py-2 text-xs font-semibold text-muted-foreground/60 select-none [&:not(:first-child)]:mt-2"
-								onSelect={handleSelect}
-								onInfoClick={handleInfoClick}
+								onSelect={ms.handleSelect}
+								onInfoClick={ms.handleInfoClick}
 							/>
 						</div>
 					</div>
@@ -293,23 +157,23 @@
 			<button
 				class={cn(
 					`inline-flex cursor-pointer items-center gap-1.5 rounded-sm bg-muted-foreground/10 px-1.5 py-1 text-xs transition hover:text-foreground focus:outline-none focus-visible:ring-2 focus-visible:ring-ring focus-visible:ring-offset-2 disabled:cursor-not-allowed disabled:opacity-60`,
-					!isCurrentModelInCache
+					!ms.isCurrentModelInCache
 						? 'bg-red-400/10 !text-red-400 hover:bg-red-400/20 hover:text-red-400'
 						: forceForegroundText
 							? 'text-foreground'
-							: isHighlightedCurrentModelActive
+							: ms.isHighlightedCurrentModelActive
 								? 'text-foreground'
 								: 'text-muted-foreground'
 				)}
 				style="max-width: min(calc(100cqw - 6.5rem), 32rem)"
-				onclick={() => handleOpenChange(true)}
-				disabled={disabled || updating}
+				onclick={() => ms.handleOpenChange(true)}
+				disabled={disabled || ms.updating}
 			>
 				<Package class="h-3.5 w-3.5" />
 
 				<TruncatedText text={selectedOption?.model || ''} class="min-w-0 font-medium" />
 
-				{#if updating}
+				{#if ms.updating}
 					<Loader2 class="h-3 w-3.5 animate-spin" />
 				{/if}
 			</button>
@@ -317,6 +181,10 @@
 	{/if}
 </div>
 
-{#if showModelDialog}
-	<DialogModelInformation bind:open={showModelDialog} modelId={infoModelId} />
+{#if ms.showModelDialog}
+	<DialogModelInformation
+		open={ms.showModelDialog}
+		onOpenChange={(v) => ms.setShowModelDialog(v)}
+		modelId={ms.infoModelId}
+	/>
 {/if}

tools/server/webui/src/lib/components/app/models/index.ts

@@ -11,7 +11,7 @@
  */
 
 /**
- * **ModelsSelector** - Model selection dropdown
+ * **ModelsSelectorDropdown** - Model selection dropdown (desktop)
  *
  * Dropdown for selecting AI models with status indicators,
  * search, and model information display. Adapts UI based on server mode.
@@ -35,20 +35,20 @@
  *
  * @example
  * ```svelte
- * <ModelsSelector
+ * <ModelsSelectorDropdown
  *   currentModel={conversation.modelId}
  *   onModelChange={(id, name) => updateModel(id)}
  *   useGlobalSelection
  * />
  * ```
  */
-export { default as ModelsSelector } from './ModelsSelector.svelte';
+export { default as ModelsSelectorDropdown } from './ModelsSelectorDropdown.svelte';
 
 /**
  * **ModelsSelectorList** - Grouped model options list
  *
  * Renders grouped model options (loaded, favorites, available) with section
- * headers and org subgroups. Shared between ModelsSelector and ModelsSelectorSheet
+ * headers and org subgroups. Shared between ModelsSelectorDropdown and ModelsSelectorSheet
  * to avoid template duplication.
  *
  * Accepts an optional `renderOption` snippet to customize how each option is
@@ -68,8 +68,8 @@ export { default as ModelsSelectorOption } from './ModelsSelectorOption.svelte';
 /**
  * **ModelsSelectorSheet** - Mobile model selection sheet
  *
- * Bottom sheet variant of ModelsSelector optimized for touch interaction
- * on mobile devices. Same functionality as ModelsSelector but uses Sheet UI
+ * Bottom sheet variant of ModelsSelectorDropdown optimized for touch interaction
+ * on mobile devices. Same functionality as ModelsSelectorDropdown but uses Sheet UI
  * instead of DropdownMenu.
  */
 export { default as ModelsSelectorSheet } from './ModelsSelectorSheet.svelte';

tools/server/webui/src/lib/components/app/navigation/DesktopIconStrip.svelte

@@ -0,0 +1,84 @@
+<script lang="ts">
+	import { goto } from '$app/navigation';
+	import { page } from '$app/state';
+	import { ActionIcon } from '$lib/components/app/actions';
+	import {
+		ICON_STRIP_TRANSITION_DURATION,
+		ICON_STRIP_TRANSITION_DELAY_MULTIPLIER,
+		SIDEBAR_ACTIONS_ITEMS
+	} from '$lib/constants';
+	import { TooltipSide } from '$lib/enums';
+	import { fade } from 'svelte/transition';
+	import { circIn } from 'svelte/easing';
+	import { onMount } from 'svelte';
+	import { useKeyboardShortcuts } from '$lib/hooks/use-keyboard-shortcuts.svelte';
+
+	interface Props {
+		sidebarOpen: boolean;
+		onSearchClick: () => void;
+	}
+
+	let { sidebarOpen = false, onSearchClick }: Props = $props();
+
+	const { handleKeydown } = useKeyboardShortcuts({ activateSearchMode: () => onSearchClick() });
+
+	let initialized = $state(false);
+	let showIcons = $derived(!sidebarOpen);
+
+	showIcons = false;
+
+	onMount(() => {
+		showIcons = !sidebarOpen;
+
+		setTimeout(() => {
+			initialized = true;
+		}, ICON_STRIP_TRANSITION_DELAY_MULTIPLIER * SIDEBAR_ACTIONS_ITEMS.length);
+	});
+</script>
+
+<svelte:window onkeydown={handleKeydown} />
+
+<div
+	class="hidden shrink-0 transition-[width] duration-200 ease-linear md:block {sidebarOpen
+		? 'w-0'
+		: 'w-[calc(var(--sidebar-width-icon)+1.5rem)]'}"
+></div>
+<aside
+	class="fixed top-0 bottom-0 left-0 z-10 hidden w-[calc(var(--sidebar-width-icon)+1.5rem)] flex-col items-center justify-between py-3 transition-opacity duration-200 ease-linear md:flex {sidebarOpen
+		? 'pointer-events-none opacity-0'
+		: 'opacity-100'}"
+>
+	<div class="mt-12 flex flex-col items-center gap-1">
+		{#each SIDEBAR_ACTIONS_ITEMS as item, i (item.tooltip)}
+			{@const onclick = item.route ? () => goto(item.route!) : onSearchClick}
+			{@const isActive = item.activeRouteId
+				? page.route.id === item.activeRouteId
+				: item.activeRoutePrefix
+					? !!page.route.id?.startsWith(item.activeRoutePrefix)
+					: false}
+			{#if showIcons}
+				<div
+					in:fade={{
+						duration: ICON_STRIP_TRANSITION_DURATION,
+						delay: !initialized
+							? ICON_STRIP_TRANSITION_DELAY_MULTIPLIER + i * ICON_STRIP_TRANSITION_DELAY_MULTIPLIER
+							: 0,
+						easing: circIn
+					}}
+				>
+					<ActionIcon
+						icon={item.icon}
+						tooltip={item.tooltip}
+						tooltipSide={TooltipSide.RIGHT}
+						size="lg"
+						iconSize="h-4 w-4"
+						class="h-9 w-9 rounded-full hover:bg-accent! {isActive
+							? 'bg-accent text-accent-foreground'
+							: ''}"
+						{onclick}
+					/>
+				</div>
+			{/if}
+		{/each}
+	</div>
+</aside>

tools/server/webui/src/lib/components/app/navigation/index.ts

@@ -63,3 +63,11 @@ export { default as DropdownMenuSearchable } from './DropdownMenuSearchable.svel
  * ```
  */
 export { default as DropdownMenuActions } from './DropdownMenuActions.svelte';
+
+/**
+ * **DesktopIconStrip** - Fixed icon strip for desktop sidebar
+ *
+ * Vertical icon strip shown on desktop when the sidebar is collapsed.
+ * Contains navigation shortcuts for new chat, search, MCP, import/export, and settings.
+ */
+export { default as DesktopIconStrip } from './DesktopIconStrip.svelte';

tools/server/webui/src/lib/components/app/settings/SettingsChat.svelte

@@ -0,0 +1,154 @@
+<script lang="ts">
+	import {
+		ChatSettingsFooter,
+		ChatSettingsFields,
+		ChatSettingsToolsTab
+	} from '$lib/components/app';
+	import {
+		SettingsChatDesktopSidebar,
+		SettingsChatMobileHeader
+	} from '$lib/components/app/settings';
+	import { config, settingsStore } from '$lib/stores/settings.svelte';
+	import {
+		NUMERIC_FIELDS,
+		POSITIVE_INTEGER_FIELDS,
+		SETTINGS_CHAT_SECTIONS,
+		SETTINGS_SECTION_TITLES,
+		type SettingsSection
+	} from '$lib/constants';
+	import { setMode } from 'mode-watcher';
+	import { ColorMode } from '$lib/enums/ui';
+	import { fade } from 'svelte/transition';
+	import { goto } from '$app/navigation';
+	import { page } from '$app/state';
+	import { setChatSettingsConfigContext } from '$lib/contexts';
+	import { settingsReferrer } from '$lib/stores/settings-referrer.svelte';
+
+	interface Props {
+		initialSection?: string;
+		getSectionHref?: (section: SettingsSection) => string;
+	}
+
+	let { initialSection, getSectionHref }: Props = $props();
+
+	let activeSlug = $derived(
+		initialSection ?? (page.params as Record<string, string | undefined>).section ?? 'general'
+	);
+	let currentSection = $derived(
+		SETTINGS_CHAT_SECTIONS.find((section) => section.slug === activeSlug) ||
+			SETTINGS_CHAT_SECTIONS[0]
+	);
+	let localConfig: SettingsConfigType = $state({ ...config() });
+
+	let mobileHeader: { updateCarousel: () => void } | undefined;
+
+	function handleThemeChange(newTheme: string) {
+		localConfig.theme = newTheme;
+		setMode(newTheme as ColorMode);
+	}
+
+	function handleConfigChange(key: string, value: string | boolean) {
+		localConfig[key] = value;
+	}
+
+	function handleReset() {
+		localConfig = { ...config() };
+		setMode(localConfig.theme as ColorMode);
+		mobileHeader?.updateCarousel();
+	}
+
+	function handleSave() {
+		if (localConfig.custom && typeof localConfig.custom === 'string' && localConfig.custom.trim()) {
+			try {
+				JSON.parse(localConfig.custom);
+			} catch (error) {
+				alert('Invalid JSON in custom parameters. Please check the format and try again.');
+				console.error(error);
+				return;
+			}
+		}
+
+		const processedConfig = { ...localConfig };
+
+		for (const field of NUMERIC_FIELDS) {
+			if (processedConfig[field] !== undefined && processedConfig[field] !== '') {
+				const numValue = Number(processedConfig[field]);
+				if (!isNaN(numValue)) {
+					if ((POSITIVE_INTEGER_FIELDS as readonly string[]).includes(field)) {
+						processedConfig[field] = Math.max(1, Math.round(numValue));
+					} else {
+						processedConfig[field] = numValue;
+					}
+				} else {
+					alert(`Invalid numeric value for ${field}. Please enter a valid number.`);
+					return;
+				}
+			}
+		}
+
+		settingsStore.updateMultipleConfig(processedConfig);
+		goto(settingsReferrer.url);
+	}
+
+	export function reset() {
+		localConfig = { ...config() };
+	}
+
+	setChatSettingsConfigContext({
+		get localConfig() {
+			return localConfig;
+		},
+		handleConfigChange,
+		handleThemeChange
+	});
+</script>
+
+<div
+	class="mx-auto flex h-full max-h-[100dvh] w-full flex-col overflow-y-auto md:pl-8"
+	in:fade={{ duration: 150 }}
+>
+	<div class="flex flex-1 flex-col gap-4 md:flex-row">
+		<SettingsChatDesktopSidebar
+			sections={SETTINGS_CHAT_SECTIONS}
+			isActive={(section: SettingsSection) => section.slug === activeSlug}
+			getHref={getSectionHref ?? ((section: SettingsSection) => `#/settings/chat/${section.slug}`)}
+		/>
+
+		<SettingsChatMobileHeader
+			sections={SETTINGS_CHAT_SECTIONS}
+			isActive={(section: SettingsSection) => section.slug === activeSlug}
+			getHref={getSectionHref ?? ((section: SettingsSection) => `#/settings/chat/${section.slug}`)}
+			bind:this={mobileHeader}
+		/>
+
+		<div class="mx-auto max-w-3xl flex-1">
+			<div class="space-y-6 p-4 md:p-6 md:pt-28">
+				<div class="grid">
+					<div class="mb-6 flex items-center gap-2 border-b border-border/30 pb-6 md:flex">
+						<currentSection.icon class="h-5 w-5" />
+						<h3 class="text-lg font-semibold">{currentSection.title}</h3>
+					</div>
+
+					{#if currentSection.title === SETTINGS_SECTION_TITLES.TOOLS}
+						<ChatSettingsToolsTab />
+					{:else if currentSection.fields}
+						<div class="space-y-6">
+							<ChatSettingsFields
+								fields={currentSection.fields}
+								{localConfig}
+								onConfigChange={handleConfigChange}
+								onThemeChange={handleThemeChange}
+							/>
+						</div>
+					{/if}
+				</div>
+
+				<div class="mt-8 border-t border-border/30 pt-6">
+					<p class="text-xs text-muted-foreground">Settings are saved in browser's localStorage</p>
+				</div>
+			</div>
+
+			<ChatSettingsFooter onReset={handleReset} onSave={handleSave} />
+		</div>
+	</div>
+</div>

tools/server/webui/src/lib/components/app/settings/SettingsChatDesktopSidebar.svelte

@@ -0,0 +1,49 @@
+<script lang="ts">
+	import { Settings } from '@lucide/svelte';
+	import type { SettingsSection, SettingsSectionTitle } from '$lib/constants';
+
+	interface Props {
+		sections: SettingsSection[];
+		isActive: (section: SettingsSection) => boolean;
+		getHref?: (section: SettingsSection) => string;
+		onSectionChange?: (section: SettingsSectionTitle) => void;
+	}
+
+	let { sections, isActive, getHref, onSectionChange }: Props = $props();
+</script>
+
+<div class="sticky top-0 hidden w-64 flex-col self-start bg-background pt-10 pb-4 md:flex">
+	<div class="flex items-center gap-2 pb-10">
+		<Settings class="h-6 w-6" />
+		<h1 class="text-2xl font-semibold">Settings</h1>
+	</div>
+	<nav class="space-y-1">
+		{#each sections as section (section.title)}
+			{#if getHref}
+				<a
+					class="flex w-full cursor-pointer items-center gap-3 rounded-lg px-3 py-2 text-left text-sm no-underline transition-colors hover:bg-accent {isActive(
+						section
+					)
+						? 'bg-accent text-accent-foreground'
+						: 'text-muted-foreground'}"
+					href={getHref(section)}
+				>
+					<section.icon class="h-4 w-4" />
+					<span class="ml-2">{section.title}</span>
+				</a>
+			{:else}
+				<button
+					class="flex w-full cursor-pointer items-center gap-3 rounded-lg px-3 py-2 text-left text-sm transition-colors hover:bg-accent {isActive(
+						section
+					)
+						? 'bg-accent text-accent-foreground'
+						: 'text-muted-foreground'}"
+					onclick={() => onSectionChange?.(section.title)}
+				>
+					<section.icon class="h-4 w-4" />
+					<span class="ml-2">{section.title}</span>
+				</button>
+			{/if}
+		{/each}
+	</nav>
+</div>

tools/server/webui/src/lib/components/app/settings/SettingsChatMobileHeader.svelte

@@ -0,0 +1,107 @@
+<script lang="ts">
+	import { Settings, ChevronLeft, ChevronRight } from '@lucide/svelte';
+	import { onMount, tick } from 'svelte';
+	import type { SettingsSection, SettingsSectionTitle } from '$lib/constants';
+	import { useScrollCarousel } from '$lib/hooks/use-scroll-carousel.svelte';
+
+	interface Props {
+		sections: SettingsSection[];
+		isActive: (section: SettingsSection) => boolean;
+		getHref?: (section: SettingsSection) => string;
+		onSectionChange?: (section: SettingsSectionTitle) => void;
+	}
+
+	let { sections, isActive, getHref, onSectionChange }: Props = $props();
+
+	const carousel = useScrollCarousel();
+
+	onMount(async () => {
+		await tick();
+		if (carousel.scrollContainer) {
+			const activeTab = carousel.scrollContainer.querySelector('[data-active="true"]');
+			if (activeTab instanceof HTMLElement) {
+				carousel.scrollToCenter(activeTab);
+			}
+		}
+	});
+
+	export function updateCarousel() {
+		setTimeout(carousel.updateScrollButtons, 100);
+	}
+</script>
+
+<div class="sticky top-0 z-10 flex flex-col bg-background md:hidden">
+	<div class="flex items-center gap-2 px-4 pt-4 pb-2 md:pt-6">
+		<Settings class="h-5 w-5 md:h-6 md:w-6" />
+
+		<h1 class="text-xl font-semibold md:text-2xl">Settings</h1>
+	</div>
+
+	<div class="border-b border-border/30 py-2">
+		<div class="relative flex items-center" style="scroll-padding: 1rem;">
+			<button
+				class="absolute left-2 z-10 flex h-6 w-6 items-center justify-center rounded-full bg-muted shadow-md backdrop-blur-sm transition-opacity hover:bg-accent {carousel.canScrollLeft
+					? 'opacity-100'
+					: 'pointer-events-none opacity-0'}"
+				onclick={carousel.scrollLeft}
+				aria-label="Scroll left"
+			>
+				<ChevronLeft class="h-4 w-4" />
+			</button>
+
+			<div
+				class="scrollbar-hide overflow-x-auto py-2"
+				bind:this={carousel.scrollContainer}
+				onscroll={carousel.updateScrollButtons}
+			>
+				<div class="flex min-w-max gap-2">
+					{#each sections as section (section.title)}
+						{#if getHref}
+							<a
+								class="flex cursor-pointer items-center gap-2 rounded-lg px-3 py-2 text-sm whitespace-nowrap no-underline transition-colors first:ml-4 last:mr-4 hover:bg-accent {isActive(
+									section
+								)
+									? 'bg-accent text-accent-foreground'
+									: 'text-muted-foreground'}"
+								data-active={isActive(section)}
+								href={getHref(section)}
+								onclick={(e: MouseEvent) => {
+									carousel.scrollToCenter(e.currentTarget as HTMLElement);
+								}}
+							>
+								<section.icon class="h-4 w-4 flex-shrink-0" />
+								<span>{section.title}</span>
+							</a>
+						{:else}
+							<button
+								class="flex cursor-pointer items-center gap-2 rounded-lg px-3 py-2 text-sm whitespace-nowrap transition-colors first:ml-4 last:mr-4 hover:bg-accent {isActive(
+									section
+								)
+									? 'bg-accent text-accent-foreground'
+									: 'text-muted-foreground'}"
+								data-active={isActive(section)}
+								onclick={(e: MouseEvent) => {
+									onSectionChange?.(section.title);
+									carousel.scrollToCenter(e.currentTarget as HTMLElement);
+								}}
+							>
+								<section.icon class="h-4 w-4 flex-shrink-0" />
+								<span>{section.title}</span>
+							</button>
+						{/if}
+					{/each}
+				</div>
+			</div>
+
+			<button
+				class="absolute right-2 z-10 flex h-6 w-6 items-center justify-center rounded-full bg-muted shadow-md backdrop-blur-sm transition-opacity hover:bg-accent {carousel.canScrollRight
+					? 'opacity-100'
+					: 'pointer-events-none opacity-0'}"
+				onclick={carousel.scrollRight}
+				aria-label="Scroll right"
+			>
+				<ChevronRight class="h-4 w-4" />
+			</button>
+		</div>
+	</div>
+</div>

tools/server/webui/src/lib/components/app/settings/SettingsImportExport.svelte

@@ -1,11 +1,21 @@
 <script lang="ts">
-	import { Download, Upload, Trash2 } from '@lucide/svelte';
-	import { Button } from '$lib/components/ui/button';
+	import type { Component } from 'svelte';
+	import { Download, Upload, Trash2, Database } from '@lucide/svelte';
+	import { Button, type ButtonVariant } from '$lib/components/ui/button';
 	import { DialogConversationSelection, DialogConfirmation } from '$lib/components/app';
 	import { createMessageCountMap } from '$lib/utils';
-	import { ISO_DATE_TIME_SEPARATOR } from '$lib/constants';
 	import { conversationsStore, conversations } from '$lib/stores/conversations.svelte';
 	import { toast } from 'svelte-sonner';
+	import { fade } from 'svelte/transition';
+	import { ConversationSelectionMode, HtmlInputType, FileExtensionText } from '$lib/enums';
+
+	interface SectionOpts {
+		wrapperClass?: string;
+		titleClass?: string;
+		buttonVariant?: ButtonVariant;
+		buttonClass?: string;
+		summary?: { show: boolean; verb: string; items: DatabaseConversation[] };
+	}
 
 	let exportedConversations = $state<DatabaseConversation[]>([]);
 	let importedConversations = $state<DatabaseConversation[]>([]);
@@ -56,10 +66,7 @@
 				})
 			);
 
-			conversationsStore.downloadConversationFile(
-				allData,
-				`${new Date().toISOString().split(ISO_DATE_TIME_SEPARATOR)[0]}_conversations.json`
-			);
+			conversationsStore.downloadConversationFile(allData);
 
 			exportedConversations = selectedConversations;
 			showExportSummary = true;
@@ -75,8 +82,8 @@
 		try {
 			const input = document.createElement('input');
 
-			input.type = 'file';
-			input.accept = '.json';
+			input.type = HtmlInputType.FILE;
+			input.accept = FileExtensionText.JSON;
 
 			input.onchange = async (e) => {
 				const file = (e.target as HTMLInputElement)?.files?.[0];
@@ -174,114 +181,104 @@
 	}
 </script>
 
-<div class="space-y-6">
-	<div class="space-y-4">
-		<div class="grid">
-			<h4 class="mb-2 text-sm font-medium">Export Conversations</h4>
+{#snippet summaryList(show: boolean, verb: string, items: DatabaseConversation[])}
+	{#if show && items.length > 0}
+		<div class="mt-4 grid overflow-x-auto rounded-lg border border-border/50 bg-muted/30 p-4">
+			<h5 class="mb-2 text-sm font-medium">
+				{verb}
+				{items.length} conversation{items.length === 1 ? '' : 's'}
+			</h5>
 
-			<p class="mb-4 text-sm text-muted-foreground">
-				Download all your conversations as a JSON file. This includes all messages, attachments, and
-				conversation history.
-			</p>
+			<ul class="space-y-1 text-sm text-muted-foreground">
+				{#each items.slice(0, 10) as conv (conv.id)}
+					<li class="truncate">• {conv.name || 'Untitled conversation'}</li>
+				{/each}
 
-			<Button
-				class="w-full justify-start justify-self-start md:w-auto"
-				onclick={handleExportClick}
-				variant="outline"
-			>
-				<Download class="mr-2 h-4 w-4" />
-
-				Export conversations
-			</Button>
-
-			{#if showExportSummary && exportedConversations.length > 0}
-				<div class="mt-4 grid overflow-x-auto rounded-lg border border-border/50 bg-muted/30 p-4">
-					<h5 class="mb-2 text-sm font-medium">
-						Exported {exportedConversations.length} conversation{exportedConversations.length === 1
-							? ''
-							: 's'}
-					</h5>
-
-					<ul class="space-y-1 text-sm text-muted-foreground">
-						{#each exportedConversations.slice(0, 10) as conv (conv.id)}
-							<li class="truncate">• {conv.name || 'Untitled conversation'}</li>
-						{/each}
-
-						{#if exportedConversations.length > 10}
-							<li class="italic">
-								... and {exportedConversations.length - 10} more
-							</li>
-						{/if}
-					</ul>
-				</div>
-			{/if}
+				{#if items.length > 10}
+					<li class="italic">... and {items.length - 10} more</li>
+				{/if}
+			</ul>
 		</div>
+	{/if}
+{/snippet}
 
-		<div class="grid border-t border-border/30 pt-4">
-			<h4 class="mb-2 text-sm font-medium">Import Conversations</h4>
+{#snippet section(
+	title: string,
+	description: string,
+	Icon: Component,
+	buttonText: string,
+	onclick: () => void,
+	opts: SectionOpts
+)}
+	{@const buttonClass = opts?.buttonClass ?? 'justify-start justify-self-start md:w-auto'}
+	{@const buttonVariant = opts?.buttonVariant ?? 'outline'}
+	<div class="grid gap-1 {opts?.wrapperClass ?? ''}">
+		<h4 class="mt-0 mb-2 text-sm font-medium {opts?.titleClass ?? ''}">{title}</h4>
 
-			<p class="mb-4 text-sm text-muted-foreground">
-				Import one or more conversations from a previously exported JSON file. This will merge with
-				your existing conversations.
-			</p>
+		<p class="mb-4 text-sm text-muted-foreground">{description}</p>
 
-			<Button
-				class="w-full justify-start justify-self-start md:w-auto"
-				onclick={handleImportClick}
-				variant="outline"
-			>
-				<Upload class="mr-2 h-4 w-4" />
-				Import conversations
-			</Button>
+		<Button class={buttonClass} {onclick} variant={buttonVariant}>
+			<Icon class="mr-2 h-4 w-4" />
 
-			{#if showImportSummary && importedConversations.length > 0}
-				<div class="mt-4 grid overflow-x-auto rounded-lg border border-border/50 bg-muted/30 p-4">
-					<h5 class="mb-2 text-sm font-medium">
-						Imported {importedConversations.length} conversation{importedConversations.length === 1
-							? ''
-							: 's'}
-					</h5>
+			{buttonText}
+		</Button>
 
-					<ul class="space-y-1 text-sm text-muted-foreground">
-						{#each importedConversations.slice(0, 10) as conv (conv.id)}
-							<li class="truncate">• {conv.name || 'Untitled conversation'}</li>
-						{/each}
+		{#if opts?.summary}
+			{@render summaryList(opts.summary.show, opts.summary.verb, opts.summary.items)}
+		{/if}
+	</div>
+{/snippet}
 
-						{#if importedConversations.length > 10}
-							<li class="italic">
-								... and {importedConversations.length - 10} more
-							</li>
-						{/if}
-					</ul>
-				</div>
-			{/if}
-		</div>
+<div class="space-y-6" in:fade={{ duration: 150 }}>
+	<div class="flex items-center gap-2 pb-4">
+		<Database class="h-5 w-5 md:h-6 md:w-6" />
 
-		<div class="grid border-t border-border/30 pt-4">
-			<h4 class="mb-2 text-sm font-medium text-destructive">Delete All Conversations</h4>
+		<h1 class="text-xl font-semibold md:text-2xl">Import / Export</h1>
+	</div>
 
-			<p class="mb-4 text-sm text-muted-foreground">
-				Permanently delete all conversations and their messages. This action cannot be undone.
-				Consider exporting your conversations first if you want to keep a backup.
-			</p>
+	<div class="space-y-6">
+		{@render section(
+			'Export Conversations',
+			'Download all your conversations as a JSON file. This includes all messages, attachments, and conversation history.',
+			Download,
+			'Export conversations',
+			handleExportClick,
+			{ summary: { show: showExportSummary, verb: 'Exported', items: exportedConversations } }
+		)}
 
-			<Button
-				class="text-destructive-foreground w-full justify-start justify-self-start bg-destructive hover:bg-destructive/80 md:w-auto"
-				onclick={handleDeleteAllClick}
-				variant="destructive"
-			>
-				<Trash2 class="mr-2 h-4 w-4" />
+		{@render section(
+			'Import Conversations',
+			'Import one or more conversations from a previously exported JSON file. This will merge with your existing conversations.',
+			Upload,
+			'Import conversations',
+			handleImportClick,
+			{
+				wrapperClass: 'border-t border-border/30 pt-6',
+				summary: { show: showImportSummary, verb: 'Imported', items: importedConversations }
+			}
+		)}
 
-				Delete all conversations
-			</Button>
-		</div>
+		{@render section(
+			'Delete All Conversations',
+			'Permanently delete all conversations and their messages. This action cannot be undone. Consider exporting your conversations first if you want to keep a backup.',
+			Trash2,
+			'Delete all conversations',
+			handleDeleteAllClick,
+			{
+				wrapperClass: 'border-t border-border/30 pt-4',
+				titleClass: 'text-destructive',
+				buttonVariant: 'destructive',
+				buttonClass:
+					'text-destructive-foreground justify-start justify-self-start bg-destructive hover:bg-destructive/80 md:w-auto'
+			}
+		)}
 	</div>
 </div>
 
 <DialogConversationSelection
 	conversations={availableConversations}
 	{messageCountMap}
-	mode="export"
+	mode={ConversationSelectionMode.EXPORT}
 	bind:open={showExportDialog}
 	onCancel={() => (showExportDialog = false)}
 	onConfirm={handleExportConfirm}
@@ -290,7 +287,7 @@
 <DialogConversationSelection
 	conversations={availableConversations}
 	{messageCountMap}
-	mode="import"
+	mode={ConversationSelectionMode.IMPORT}
 	bind:open={showImportDialog}
 	onCancel={() => (showImportDialog = false)}
 	onConfirm={handleImportConfirm}

tools/server/webui/src/lib/components/app/settings/SettingsMcpServers.svelte

@@ -0,0 +1,109 @@
+<script lang="ts">
+	import { Plus } from '@lucide/svelte';
+	import { Button } from '$lib/components/ui/button';
+	import { mcpStore } from '$lib/stores/mcp.svelte';
+	import { conversationsStore } from '$lib/stores/conversations.svelte';
+	import { toolsStore } from '$lib/stores/tools.svelte';
+	import { McpServerCard, McpServerCardSkeleton } from '$lib/components/app/mcp';
+	import { DialogMcpServerAddNew } from '$lib/components/app/dialogs';
+	import { HealthCheckStatus } from '$lib/enums';
+	import { fade } from 'svelte/transition';
+	import { onMount } from 'svelte';
+	import McpLogo from '../mcp/McpLogo.svelte';
+	import { page } from '$app/state';
+	import { replaceState } from '$app/navigation';
+
+	interface Props {
+		class?: string;
+	}
+
+	let { class: className }: Props = $props();
+
+	let servers = $derived(mcpStore.getServersSorted());
+
+	let initialLoadComplete = $state(false);
+	let isAddingServer = $state(false);
+
+	onMount(() => {
+		if (page.url.searchParams.has('add')) {
+			isAddingServer = true;
+
+			const newUrl = new URL(page.url);
+			newUrl.searchParams.delete('add');
+
+			replaceState(newUrl, {});
+		}
+	});
+
+	$effect(() => {
+		if (initialLoadComplete) return;
+
+		const allChecked =
+			servers.length > 0 &&
+			servers.every((server) => {
+				const state = mcpStore.getHealthCheckState(server.id);
+
+				return (
+					state.status === HealthCheckStatus.SUCCESS || state.status === HealthCheckStatus.ERROR
+				);
+			});
+
+		if (allChecked) {
+			initialLoadComplete = true;
+		}
+	});
+</script>
+
+<div in:fade={{ duration: 150 }} class="max-h-full overflow-auto">
+	<div class="flex items-center gap-2 p-4 md:absolute md:top-8 md:left-8 md:px-0 md:py-2">
+		<McpLogo class="h-5 w-5 md:h-6 md:w-6" />
+
+		<h1 class="text-xl font-semibold md:text-2xl">MCP Servers</h1>
+	</div>
+
+	<div class="sticky top-0 z-10 mt-4 flex items-start justify-end gap-4 px-8 py-4">
+		<Button variant="outline" size="sm" class="shrink-0" onclick={() => (isAddingServer = true)}>
+			<Plus class="h-4 w-4" />
+
+			Add New Server
+		</Button>
+	</div>
+
+	<DialogMcpServerAddNew bind:open={isAddingServer} />
+
+	<div class="grid gap-5 md:space-y-4 {className}">
+		{#if servers.length === 0 && !isAddingServer}
+			<div class="rounded-md border border-dashed p-4 text-sm text-muted-foreground">
+				No MCP Servers configured yet. Add one to enable agentic features.
+			</div>
+		{/if}
+
+		{#if servers.length > 0}
+			<div
+				class="grid gap-3"
+				style="grid-template-columns: repeat(auto-fill, minmax(min(32rem, calc(100dvw - 2rem)), 1fr));"
+			>
+				{#each servers as server (server.id)}
+					{#if !initialLoadComplete}
+						<McpServerCardSkeleton />
+					{:else}
+						<McpServerCard
+							{server}
+							faviconUrl={mcpStore.getServerFavicon(server.id)}
+							enabled={conversationsStore.isMcpServerEnabledForChat(server.id)}
+							onToggle={async () => {
+								const wasEnabled = conversationsStore.isMcpServerEnabledForChat(server.id);
+								await conversationsStore.toggleMcpServerForChat(server.id);
+								if (!wasEnabled) {
+									toolsStore.enableAllToolsForServer(server.id);
+								}
+							}}
+							onUpdate={(updates) => mcpStore.updateServer(server.id, updates)}
+							onDelete={() => mcpStore.removeServer(server.id)}
+						/>
+					{/if}
+				{/each}
+			</div>
+		{/if}
+	</div>
+</div>

tools/server/webui/src/lib/components/app/settings/index.ts

@@ -0,0 +1,32 @@
+/**
+ * Full chat settings page layout with sidebar, mobile header, and content area.
+ * Manages local configuration state, section navigation, and context setup.
+ * Accepts an optional `initialSection` prop to override the URL-based section resolution.
+ */
+export { default as SettingsChat } from './SettingsChat.svelte';
+
+/**
+ * Desktop sidebar navigation for chat settings.
+ * Displays a list of settings sections with icons and titles.
+ * Supports both hash-link navigation (via `getHref`) and in-app section switching (via `onSectionChange`).
+ */
+export { default as SettingsChatDesktopSidebar } from './SettingsChatDesktopSidebar.svelte';
+
+/**
+ * Mobile header with a horizontally scrollable section picker for chat settings.
+ * Shows chevron buttons for scroll navigation and highlights the active section.
+ * Supports both hash-link navigation (via `getHref`) and in-app section switching (via `onSectionChange`).
+ */
+export { default as SettingsChatMobileHeader } from './SettingsChatMobileHeader.svelte';
+
+/**
+ * Settings Import/Export panel.
+ * Provides UI for importing and exporting chat settings configurations.
+ */
+export { default as SettingsImportExport } from './SettingsImportExport.svelte';
+
+/**
+ * MCP Servers configuration panel.
+ * Provides UI for managing Model Context Protocol (MCP) server connections.
+ */
+export { default as SettingsMcpServers } from './SettingsMcpServers.svelte';

tools/server/webui/src/lib/components/ui/button-group/button-group-root.svelte

@@ -0,0 +1,21 @@
+<script lang="ts">
+	import { cn } from '$lib/components/ui/utils';
+	import type { Snippet } from 'svelte';
+	import type { HTMLAttributes } from 'svelte/elements';
+
+	interface Props extends HTMLAttributes<HTMLDivElement> {
+		children: Snippet;
+	}
+
+	let { class: className, children, ...restProps }: Props = $props();
+</script>
+
+<div
+	class={cn(
+		'flex items-center [&>*:first-child]:rounded-r-none [&>*:last-child]:rounded-l-none [&>*:not(:first-child):not(:last-child)]:rounded-none',
+		className
+	)}
+	{...restProps}
+>
+	{@render children()}
+</div>

tools/server/webui/src/lib/components/ui/button-group/button-group-separator.svelte

@@ -0,0 +1,8 @@
+<script lang="ts">
+	import { cn } from '$lib/components/ui/utils';
+	import type { HTMLAttributes } from 'svelte/elements';
+
+	let { ...restProps }: HTMLAttributes<HTMLDivElement> = $props();
+</script>
+
+<div class={cn('shrink-0 self-stretch bg-border', 'w-px')} {...restProps}></div>

tools/server/webui/src/lib/components/ui/button-group/index.ts

@@ -0,0 +1,2 @@
+export { default as Root } from './button-group-root.svelte';
+export { default as Separator } from './button-group-separator.svelte';

tools/server/webui/src/lib/components/ui/button/button.svelte

@@ -9,9 +9,9 @@
 			variant: {
 				default: 'bg-primary text-primary-foreground shadow-xs hover:bg-primary/90',
 				destructive:
-					'bg-destructive shadow-xs hover:bg-destructive/90 focus-visible:ring-destructive/20 dark:focus-visible:ring-destructive/40 dark:bg-destructive/60 text-white',
+					'bg-destructive shadow-xs hover:bg-destructive/90 focus-visible:ring-destructive/20 dark:focus-visible:ring-destructive/40 dark:bg-destructive/60 text-white!',
 				outline:
-					'bg-background shadow-xs hover:bg-accent hover:text-accent-foreground dark:bg-input/30 dark:border-input dark:hover:bg-input/50 border',
+					'shadow-xs hover:text-accent-foreground hover:bg-muted-foreground/10 backdrop-blur-sm dark:border-input border',
 				secondary:
 					'dark:bg-secondary dark:text-secondary-foreground bg-background shadow-sm text-foreground hover:bg-muted-foreground/20',
 				ghost: 'hover:text-accent-foreground hover:bg-muted-foreground/10 backdrop-blur-sm',

tools/server/webui/src/lib/components/ui/dropdown-menu/dropdown-menu-sub-content.svelte

@@ -13,7 +13,7 @@
 	bind:ref
 	data-slot="dropdown-menu-sub-content"
 	class={cn(
-		'z-50 min-w-[8rem] origin-(--bits-dropdown-menu-content-transform-origin) overflow-hidden rounded-md border bg-popover p-1 text-popover-foreground shadow-lg data-[side=bottom]:slide-in-from-top-2 data-[side=left]:slide-in-from-right-2 data-[side=right]:slide-in-from-left-2 data-[side=top]:slide-in-from-bottom-2 data-[state=closed]:animate-out data-[state=closed]:fade-out-0 data-[state=closed]:zoom-out-95 data-[state=open]:animate-in data-[state=open]:fade-in-0 data-[state=open]:zoom-in-95',
+		'z-50 max-h-(--bits-dropdown-menu-content-available-height) min-w-[8rem] origin-(--bits-dropdown-menu-content-transform-origin) overflow-x-hidden overflow-y-auto rounded-md border border-border bg-popover p-1.5 text-popover-foreground shadow-md outline-none data-[side=bottom]:slide-in-from-top-2 data-[side=left]:slide-in-from-right-2 data-[side=right]:slide-in-from-left-2 data-[side=top]:slide-in-from-bottom-2 data-[state=closed]:animate-out data-[state=closed]:fade-out-0 data-[state=closed]:zoom-out-95 data-[state=open]:animate-in data-[state=open]:fade-in-0 data-[state=open]:zoom-in-95 dark:border-border/20',
 		className
 	)}
 	{...restProps}

tools/server/webui/src/lib/components/ui/sidebar/constants.ts

@@ -1,6 +1,7 @@
 export const SIDEBAR_COOKIE_NAME = 'sidebar:state';
 export const SIDEBAR_COOKIE_MAX_AGE = 60 * 60 * 24 * 7;
-export const SIDEBAR_WIDTH = '18rem';
+export const SIDEBAR_MIN_WIDTH = '18rem';
+export const SIDEBAR_MAX_WIDTH = '32rem';
 export const SIDEBAR_WIDTH_MOBILE = '18rem';
 export const SIDEBAR_WIDTH_ICON = '3rem';
 export const SIDEBAR_KEYBOARD_SHORTCUT = 'b';

tools/server/webui/src/lib/components/ui/sidebar/context.svelte.ts

@@ -1,6 +1,6 @@
 import { IsMobile } from '$lib/hooks/is-mobile.svelte.js';
 import { getContext, setContext } from 'svelte';
-import { SIDEBAR_KEYBOARD_SHORTCUT } from './constants.js';
+import { SIDEBAR_KEYBOARD_SHORTCUT, SIDEBAR_MIN_WIDTH } from './constants.js';
 
 type Getter<T> = () => T;
 
@@ -24,6 +24,8 @@ class SidebarState {
 	readonly props: SidebarStateProps;
 	open = $derived.by(() => this.props.open());
 	openMobile = $state(false);
+	sidebarWidth = $state(SIDEBAR_MIN_WIDTH);
+	isResizing = $state(false);
 	setOpen: SidebarStateProps['setOpen'];
 	#isMobile: IsMobile;
 	state = $derived.by(() => (this.open ? 'expanded' : 'collapsed'));
@@ -53,7 +55,7 @@ class SidebarState {
 	};
 
 	toggle = () => {
-		return this.#isMobile.current ? (this.openMobile = !this.openMobile) : this.setOpen(!this.open);
+		this.setOpen(!this.open);
 	};
 }
 

tools/server/webui/src/lib/components/ui/sidebar/sidebar-footer.svelte

@@ -14,7 +14,7 @@
 	bind:this={ref}
 	data-slot="sidebar-footer"
 	data-sidebar="footer"
-	class={cn('flex flex-col gap-2 p-2', className)}
+	class={cn('flex flex-col gap-2 p-3', className)}
 	{...restProps}
 >
 	{@render children?.()}

tools/server/webui/src/lib/components/ui/sidebar/sidebar-menu-button.svelte

@@ -2,7 +2,7 @@
 	import { tv, type VariantProps } from 'tailwind-variants';
 
 	export const sidebarMenuButtonVariants = tv({
-		base: 'peer/menu-button outline-hidden ring-sidebar-ring hover:bg-sidebar-accent hover:text-sidebar-accent-foreground active:bg-sidebar-accent active:text-sidebar-accent-foreground group-has-data-[sidebar=menu-action]/menu-item:pr-8 data-[active=true]:bg-sidebar-accent data-[active=true]:text-sidebar-accent-foreground data-[state=open]:hover:bg-sidebar-accent data-[state=open]:hover:text-sidebar-accent-foreground group-data-[collapsible=icon]:size-8! group-data-[collapsible=icon]:p-2! flex w-full items-center gap-2 overflow-hidden rounded-md p-2 text-left text-sm transition-[width,height,padding] focus-visible:ring-2 disabled:pointer-events-none disabled:opacity-50 aria-disabled:pointer-events-none aria-disabled:opacity-50 data-[active=true]:font-medium [&>span:last-child]:truncate [&>svg]:size-4 [&>svg]:shrink-0',
+		base: 'peer/menu-button outline-hidden ring-sidebar-ring hover:bg-sidebar-accent hover:text-sidebar-accent-foreground active:bg-sidebar-accent active:text-sidebar-accent-foreground group-has-data-[sidebar=menu-action]/menu-item:pr-8 data-[active=true]:bg-sidebar-accent data-[active=true]:text-sidebar-accent-foreground data-[state=open]:hover:bg-sidebar-accent data-[state=open]:hover:text-sidebar-accent-foreground group-data-[collapsible=icon]:size-8! group-data-[collapsible=icon]:p-2! flex w-full items-center gap-2 overflow-hidden rounded-md py-2 px-1 text-left text-sm transition-[width,height,padding] focus-visible:ring-2 disabled:pointer-events-none disabled:opacity-50 aria-disabled:pointer-events-none aria-disabled:opacity-50 data-[active=true]:font-medium [&>span:last-child]:truncate [&>svg]:size-4 [&>svg]:shrink-0',
 		variants: {
 			variant: {
 				default: 'hover:bg-sidebar-accent hover:text-sidebar-accent-foreground',

tools/server/webui/src/lib/components/ui/sidebar/sidebar-provider.svelte

@@ -4,7 +4,8 @@
 	import {
 		SIDEBAR_COOKIE_MAX_AGE,
 		SIDEBAR_COOKIE_NAME,
-		SIDEBAR_WIDTH,
+		SIDEBAR_MIN_WIDTH,
+		SIDEBAR_MAX_WIDTH,
 		SIDEBAR_WIDTH_ICON
 	} from './constants.js';
 	import { setSidebar } from './context.svelte.js';
@@ -38,7 +39,7 @@
 
 <div
 	data-slot="sidebar-wrapper"
-	style="--sidebar-width: {SIDEBAR_WIDTH}; --sidebar-width-icon: {SIDEBAR_WIDTH_ICON}; {style}"
+	style="--sidebar-width: {sidebar.sidebarWidth}; --sidebar-min-width: {SIDEBAR_MIN_WIDTH}; --sidebar-max-width: {SIDEBAR_MAX_WIDTH}; --sidebar-width-icon: {SIDEBAR_WIDTH_ICON}; {style}"
 	class={cn(
 		'group/sidebar-wrapper flex min-h-svh w-full has-data-[variant=inset]:bg-sidebar',
 		className

tools/server/webui/src/lib/components/ui/sidebar/sidebar-trigger.svelte

@@ -3,6 +3,7 @@
 	import PanelLeftIcon from '@lucide/svelte/icons/panel-left';
 	import type { ComponentProps } from 'svelte';
 	import { useSidebar } from './context.svelte.js';
+	import { PanelLeftClose } from '@lucide/svelte';
 
 	let {
 		ref = $bindable(null),
@@ -21,9 +22,11 @@
 	data-slot="sidebar-trigger"
 	variant="ghost"
 	size="icon-lg"
-	class="rounded-full backdrop-blur-lg {className} md:left-{sidebar.open
-		? 'unset'
-		: '2'} -top-2 -left-2 md:top-0"
+	class="rounded-full backdrop-blur-lg {className} {sidebar.open
+		? 'top-1.5'
+		: 'top-0'} md:left-[calc(var(--sidebar-width)-3.25rem)] {sidebar.isResizing
+		? '!duration-0'
+		: ''}"
 	type="button"
 	onclick={(e) => {
 		onclick?.(e);
@@ -31,6 +34,10 @@
 	}}
 	{...restProps}
 >
-	<PanelLeftIcon />
+	{#if sidebar.open}
+		<PanelLeftClose />
+	{:else}
+		<PanelLeftIcon />
+	{/if}
 	<span class="sr-only">Toggle Sidebar</span>
 </Button>

tools/server/webui/src/lib/components/ui/sidebar/sidebar.svelte

@@ -1,9 +1,9 @@
 <script lang="ts">
-	import * as Sheet from '$lib/components/ui/sheet/index.js';
 	import { cn, type WithElementRef } from '$lib/components/ui/utils.js';
 	import type { HTMLAttributes } from 'svelte/elements';
-	import { SIDEBAR_WIDTH_MOBILE } from './constants.js';
+	import { SIDEBAR_MIN_WIDTH, SIDEBAR_MAX_WIDTH } from './constants.js';
 	import { useSidebar } from './context.svelte.js';
+	import { remToPx } from '$lib/utils';
 
 	let {
 		ref = $bindable(null),
@@ -20,6 +20,34 @@
 	} = $props();
 
 	const sidebar = useSidebar();
+
+	function handleResizePointerDown(e: PointerEvent) {
+		if (sidebar.isMobile) return;
+		e.preventDefault();
+
+		const target = e.currentTarget as HTMLElement;
+		target.setPointerCapture(e.pointerId);
+
+		const minPx = remToPx(SIDEBAR_MIN_WIDTH);
+		const maxPx = remToPx(SIDEBAR_MAX_WIDTH);
+
+		sidebar.isResizing = true;
+
+		function onPointerMove(ev: PointerEvent) {
+			const newWidth = side === 'left' ? ev.clientX : window.innerWidth - ev.clientX;
+			const clamped = Math.min(maxPx, Math.max(minPx, newWidth));
+			sidebar.sidebarWidth = `${clamped}px`;
+		}
+
+		function onPointerUp() {
+			sidebar.isResizing = false;
+			target.removeEventListener('pointermove', onPointerMove);
+			target.removeEventListener('pointerup', onPointerUp);
+		}
+
+		target.addEventListener('pointermove', onPointerMove);
+		target.addEventListener('pointerup', onPointerUp);
+	}
 </script>
 
 {#if collapsible === 'none'}
@@ -33,29 +61,10 @@
 	>
 		{@render children?.()}
 	</div>
-{:else if sidebar.isMobile}
-	<Sheet.Root bind:open={() => sidebar.openMobile, (v) => sidebar.setOpenMobile(v)} {...restProps}>
-		<Sheet.Content
-			data-sidebar="sidebar"
-			data-slot="sidebar"
-			data-mobile="true"
-			class="z-99999 w-(--sidebar-width) bg-sidebar p-0 text-sidebar-foreground sm:z-99 [&>button]:hidden"
-			style="--sidebar-width: {SIDEBAR_WIDTH_MOBILE};"
-			{side}
-		>
-			<Sheet.Header class="sr-only">
-				<Sheet.Title>Sidebar</Sheet.Title>
-				<Sheet.Description>Displays the mobile sidebar.</Sheet.Description>
-			</Sheet.Header>
-			<div class="flex h-full w-full flex-col">
-				{@render children?.()}
-			</div>
-		</Sheet.Content>
-	</Sheet.Root>
 {:else}
 	<div
 		bind:this={ref}
-		class="group peer hidden text-sidebar-foreground md:block"
+		class="group peer block text-sidebar-foreground"
 		data-state={sidebar.state}
 		data-collapsible={sidebar.state === 'collapsed' ? collapsible : ''}
 		data-variant={variant}
@@ -66,36 +75,76 @@
 		<div
 			data-slot="sidebar-gap"
 			class={cn(
-				'relative w-(--sidebar-width) bg-transparent transition-[width] duration-200 ease-linear',
-				'group-data-[collapsible=offcanvas]:w-0',
+				'relative bg-transparent transition-[width] duration-200 ease-linear',
+				sidebar.isResizing && '!duration-0',
+				'w-0',
+				variant === 'floating'
+					? 'md:w-[calc(var(--sidebar-width)+0.75rem)]'
+					: 'md:w-(--sidebar-width)',
+				'md:group-data-[collapsible=offcanvas]:w-0',
 				'group-data-[side=right]:rotate-180',
 				variant === 'floating' || variant === 'inset'
 					? 'group-data-[collapsible=icon]:w-[calc(var(--sidebar-width-icon)+(--spacing(4))+2px)]'
 					: 'group-data-[collapsible=icon]:w-(--sidebar-width-icon)'
 			)}
 		></div>
+
 		<div
 			data-slot="sidebar-container"
 			class={cn(
-				'fixed inset-y-0 z-999 hidden h-svh w-(--sidebar-width) transition-[left,right,width] duration-200 ease-linear md:z-0 md:flex',
-				side === 'left'
-					? 'left-0 group-data-[collapsible=offcanvas]:left-[calc(var(--sidebar-width)*-1)]'
-					: 'right-0 group-data-[collapsible=offcanvas]:right-[calc(var(--sidebar-width)*-1)]',
-				// Adjust the padding for floating and inset variants.
-				variant === 'floating' || variant === 'inset'
+				'fixed inset-y-0 z-[900] flex w-[calc(100dvw-1.5rem)] duration-200 ease-linear md:z-0 md:w-(--sidebar-width)',
+				'group-data-[collapsible=offcanvas]:pointer-events-none md:group-data-[collapsible=offcanvas]:pointer-events-auto',
+				sidebar.isResizing && '!duration-0',
+				variant === 'floating'
+					? [
+							'transition-[left,right,width,opacity]',
+							side === 'left'
+								? 'left-3 group-data-[collapsible=offcanvas]:left-[calc(var(--sidebar-width)*-0.775)] group-data-[collapsible=offcanvas]:opacity-0'
+								: 'right-3 group-data-[collapsible=offcanvas]:right-[calc(var(--sidebar-width)*-0.775)] group-data-[collapsible=offcanvas]:opacity-0',
+							'my-3 overflow-hidden rounded-3xl border border-sidebar-border shadow-md'
+						]
+					: [
+							'h-svh transition-[left,right,width]',
+							side === 'left'
+								? 'left-0 group-data-[collapsible=offcanvas]:left-[calc(var(--sidebar-width)*-1)]'
+								: 'right-0 group-data-[collapsible=offcanvas]:right-[calc(var(--sidebar-width)*-1)]'
+						],
+				// Adjust the padding for inset variant.
+				variant === 'inset'
 					? 'p-2 group-data-[collapsible=icon]:w-[calc(var(--sidebar-width-icon)+(--spacing(4))+2px)]'
-					: 'group-data-[collapsible=icon]:w-(--sidebar-width-icon)',
+					: variant === 'floating'
+						? 'group-data-[collapsible=icon]:w-[calc(var(--sidebar-width-icon)+(--spacing(4))+2px)]'
+						: 'group-data-[collapsible=icon]:w-(--sidebar-width-icon)',
 				className
 			)}
+			style={variant === 'floating' ? 'height: calc(100dvh - 1.5rem);' : undefined}
 			{...restProps}
 		>
 			<div
 				data-sidebar="sidebar"
 				data-slot="sidebar-inner"
-				class="flex h-full w-full flex-col bg-sidebar group-data-[variant=floating]:rounded-lg group-data-[variant=floating]:border group-data-[variant=floating]:border-sidebar-border group-data-[variant=floating]:shadow-sm"
+				class="flex h-full w-full flex-col bg-sidebar"
 			>
 				{@render children?.()}
 			</div>
+			<!-- Resize handle -->
+			{#if side === 'left'}
+				<!-- svelte-ignore a11y_no_static_element_interactions -->
+				<div
+					data-slot="sidebar-resize-handle"
+					class="absolute inset-y-0 right-0 z-50 hidden w-1.5 cursor-ew-resize touch-none select-none hover:bg-sidebar-border/50 active:bg-sidebar-border md:block"
+					class:bg-sidebar-border={sidebar.isResizing}
+					onpointerdown={handleResizePointerDown}
+				></div>
+			{:else}
+				<!-- svelte-ignore a11y_no_static_element_interactions -->
+				<div
+					data-slot="sidebar-resize-handle"
+					class="absolute inset-y-0 left-0 z-50 hidden w-1.5 cursor-ew-resize touch-none select-none hover:bg-sidebar-border/50 active:bg-sidebar-border md:block"
+					class:bg-sidebar-border={sidebar.isResizing}
+					onpointerdown={handleResizePointerDown}
+				></div>
+			{/if}
 		</div>
 	</div>
 {/if}

tools/server/webui/src/lib/components/ui/tooltip/tooltip-trigger.svelte

@@ -4,4 +4,9 @@
 	let { ref = $bindable(null), ...restProps }: TooltipPrimitive.TriggerProps = $props();
 </script>
 
-<TooltipPrimitive.Trigger bind:ref data-slot="tooltip-trigger" {...restProps} />
+<TooltipPrimitive.Trigger
+	bind:ref
+	data-slot="tooltip-trigger"
+	class="cursor-pointer"
+	{...restProps}
+/>

tools/server/webui/src/lib/constants/agentic.ts

@@ -4,8 +4,6 @@ export const ATTACHMENT_SAVED_REGEX = /\[Attachment saved: ([^\]]+)\]/;
 
 export const NEWLINE_SEPARATOR = '\n';
 
-export const TURN_LIMIT_MESSAGE = '\n\n```\nTurn limit reached\n```\n';
-
 export const LLM_ERROR_BLOCK_START = '\n\n```\nUpstream LLM error:\n';
 export const LLM_ERROR_BLOCK_END = '\n```\n';
 

tools/server/webui/src/lib/constants/api-endpoints.ts

@@ -4,5 +4,10 @@ export const API_MODELS = {
 	UNLOAD: '/models/unload'
 };
 
+export const API_TOOLS = {
+	LIST: '/tools',
+	EXECUTE: '/tools'
+};
+
 /** CORS proxy endpoint path */
 export const CORS_PROXY_ENDPOINT = '/cors-proxy';

tools/server/webui/src/lib/constants/attachment-menu.ts

@@ -0,0 +1,103 @@
+import type { Component } from 'svelte';
+import { MessageSquare, Zap, FolderOpen } from '@lucide/svelte';
+import { FILE_TYPE_ICONS } from '$lib/constants/icons';
+import {
+	AttachmentAction,
+	AttachmentItemEnabledWhen,
+	AttachmentItemVisibleWhen,
+	AttachmentMenuItemId
+} from '$lib/enums';
+
+export interface AttachmentMenuItem {
+	/** Unique identifier for the item */
+	id: AttachmentMenuItemId;
+	/** Display label */
+	label: string;
+	/** Lucide icon component */
+	icon: Component;
+	/** Extra CSS class applied to the item (e.g. for test selectors) */
+	class?: string;
+	/** Whether the item requires a specific modality to be enabled */
+	enabledWhen?: AttachmentItemEnabledWhen;
+	/** Tooltip shown when the item is disabled */
+	disabledTooltip?: string;
+	/** Callback key on the Props interface to invoke when clicked */
+	action: AttachmentAction;
+	/** Whether the item is only shown when a specific capability is present */
+	visibleWhen?: AttachmentItemVisibleWhen;
+	/** Whether this item has a tooltip even when enabled (uses dynamic text) */
+	hasEnabledTooltip?: boolean;
+}
+
+/**
+ * File attachment menu items shown in both the desktop dropdown and mobile sheet.
+ * The "Tools" submenu is handled separately by each component.
+ */
+export const ATTACHMENT_FILE_ITEMS: AttachmentMenuItem[] = [
+	{
+		id: AttachmentMenuItemId.IMAGES,
+		label: 'Images',
+		icon: FILE_TYPE_ICONS.image,
+		class: 'images-button',
+		enabledWhen: AttachmentItemEnabledWhen.HAS_VISION_MODALITY,
+		disabledTooltip: 'Image processing requires a vision model',
+		action: AttachmentAction.FILE_UPLOAD
+	},
+	{
+		id: AttachmentMenuItemId.AUDIO,
+		label: 'Audio Files',
+		icon: FILE_TYPE_ICONS.audio,
+		class: 'audio-button',
+		enabledWhen: AttachmentItemEnabledWhen.HAS_AUDIO_MODALITY,
+		disabledTooltip: 'Audio files processing requires an audio model',
+		action: AttachmentAction.FILE_UPLOAD
+	},
+	{
+		id: AttachmentMenuItemId.TEXT,
+		label: 'Text Files',
+		icon: FILE_TYPE_ICONS.text,
+		enabledWhen: AttachmentItemEnabledWhen.ALWAYS,
+		action: AttachmentAction.FILE_UPLOAD
+	},
+	{
+		id: AttachmentMenuItemId.PDF,
+		label: 'PDF Files',
+		icon: FILE_TYPE_ICONS.pdf,
+		enabledWhen: AttachmentItemEnabledWhen.ALWAYS,
+		disabledTooltip: 'PDFs will be converted to text. Image-based PDFs may not work properly.',
+		hasEnabledTooltip: true,
+		action: AttachmentAction.FILE_UPLOAD
+	}
+];
+
+export const ATTACHMENT_EXTRA_ITEMS: AttachmentMenuItem[] = [
+	{
+		id: AttachmentMenuItemId.SYSTEM_MESSAGE,
+		label: 'System Message',
+		icon: MessageSquare,
+		enabledWhen: AttachmentItemEnabledWhen.ALWAYS,
+		hasEnabledTooltip: true,
+		action: AttachmentAction.SYSTEM_PROMPT_CLICK
+	}
+];
+
+export const ATTACHMENT_MCP_ITEMS: AttachmentMenuItem[] = [
+	{
+		id: AttachmentMenuItemId.MCP_PROMPT,
+		label: 'MCP Prompt',
+		icon: Zap,
+		enabledWhen: AttachmentItemEnabledWhen.ALWAYS,
+		action: AttachmentAction.MCP_PROMPT_CLICK,
+		visibleWhen: AttachmentItemVisibleWhen.HAS_MCP_PROMPTS_SUPPORT
+	},
+	{
+		id: AttachmentMenuItemId.MCP_RESOURCES,
+		label: 'MCP Resources',
+		icon: FolderOpen,
+		enabledWhen: AttachmentItemEnabledWhen.ALWAYS,
+		action: AttachmentAction.MCP_RESOURCES_CLICK,
+		visibleWhen: AttachmentItemVisibleWhen.HAS_MCP_RESOURCES_SUPPORT
+	}
+];
+
+export const ATTACHMENT_TOOLTIP_TEXT = 'Add files, system prompt or MCP Servers';