541 commits

Author	SHA1	Message	Date
Haozhe	ceaa96942b	feat(kap-server): add global message search with literal and live-session modes (#2321) ... * feat(kap-server): add the /api/v1/search global message search endpoint Cross-session full-text search over user messages, assistant text and session titles, backed by a minidb index under <home>/search-index with a single-writer lock election and read-only WAL catch-up for other processes. Hits carry transcript anchors (turn ordinal and step id) so clients can jump straight to the matching turn or step. * feat(kimi-inspect): add a search view with chat-timeline navigation The left rail gains a Search view over the global search endpoint, with role and sort filters and cursor pagination. Clicking a hit switches to the chat view and navigates to its session, agent, turn and step — the channel pages the turn into the loaded window, scrolls it into view and flashes the target briefly. * chore(kimi-code): start the dev server without built web assets The repo's dev server scripts (dev:server, dev:kap-server, dev:kap-server:multi, dev:server:restart) now set KIMI_CODE_DEV_SERVER=1. When it is set and dist-web/index.html is missing, kimi web starts the API server without the bundled web UI instead of failing at startup, so backend dev no longer requires a kimi-web build. * feat(kap-server): add literal substring search and a live session route - minidb: text indexes accept an injectable tokenizer/queryTokenizer, and a hashed 2/3-gram tokenizer (NFKC + lowercase, code-point windows) backs substring search; tokenizer names persist in db.textindexes.json with backward-compatible defaults - /api/v1/search gains mode: 'literal' — n-gram candidates confirmed against the original text (zero false positives), with an 'candidate_cap' incomplete flag when the candidate set truncates - container.session_id queries against a session live in this process scan the in-memory transcript store instead of the index (both modes); the response's source: live|index field names the serving route and rides in the page-token fingerprint - kimi-inspect: exact-match toggle and source badge in the search view, plus an in-chat session search bar with jump-to-hit navigation * test(kimi-inspect): avoid stringifying BodyInit in search api tests	2026-07-28 21:47:50 +08:00
Haozhe	f79fde2b90	feat(node-sdk): migrate the SDK method surface to agent-core-v2 (#2262) ... * feat(node-sdk): add agent-core-v2 backed SDKRpcClientV2 harness - add SDKRpcClientV2 wiring the v2 engine (DI x Scope) in-process via the klient memory transport, with getExperimentalFeatures migrated to klient.global.flags.list() and unmigrated methods failing fast - export createKimiHarnessV2 / SDKRpcClientV2 from the SDK index - wire the experimental v2 gate into the CLI interactive shell (run-shell) - extend build-dts to bundle agent-core-v2 and klient declarations * feat(node-sdk): migrate listWorkspaceSkills to agent-core-v2 - add engineAccessor escape hatch exposing the in-process engine's app-scope service accessor for SDK methods the klient facade does not cover yet - implement listWorkspaceSkills via ISkillDiscovery plus the v2 user/project root helpers and BUILTIN_SKILLS (plugin skills and skillDirs still gaps) - add a v1-v2 parity test pinning identical return values per migrated method, with understood gaps listed explicitly in KNOWN_DIFFS * feat(node-sdk): migrate the SDK method surface to agent-core-v2 - implement the remaining SDKRpcClientBase methods on SDKRpcClientV2, routed through the klient facade where covered, the engineAccessor escape hatch where the engine has a service, or SDK-side rebuilds on v2 primitives where only primitives exist (config shape mapping, global mcp.json store, MCP OAuth flows, importContext, session warnings, print background policy) - translate the v2 event stream into the v1 Event union and bridge approval/question/user_tool interactions per live session - rebuild resume replay by folding the v2 wire.jsonl through the v1 agent restore pipeline, so resumed sessions render history again - keep deleteSession as not_implemented; the v2 engine has no delete capability - extend the v1-v2 parity suite to every migrated method, pinning understood engine differences in KNOWN_DIFFS - add the dev:cli:v2 root script to launch the TUI on the v2 engine * fix(node-sdk): await the v2 undo and compaction-cancel agent calls * test(cli): spread the real oauth module in the telemetry test mock * feat(node-sdk): forward v2 engine telemetry to the host telemetry client * feat(cli): gate the v2 TUI route behind a dedicated KIMI_CODE_TUI_V2 switch * fix(node-sdk): honor skillDirs on the agent-core-v2 SDK route The v2 SDK client accepted KimiHarnessOptions.skillDirs (the CLI's --skills-dir) but never seeded it into the engine, so explicit skill dirs were silently dropped on the v2 TUI route and the Skill tool could not find skills from them. Seed skillCatalogRuntimeOptions at bootstrap and let listWorkspaceSkills resolve the explicit dirs as the user source, matching the engine's session skill catalog. * refactor(cli): gate the v2 TUI route behind the master experimental flag again Drop the dedicated KIMI_CODE_TUI_V2 switch: the TUI v2 harness route is gated by KIMI_CODE_EXPERIMENTAL_FLAG, the same master switch as the kimi -p v2 route. The gate tests are kept with updated assertions, and dev:cli:v2 sets the master flag again.	2026-07-28 20:29:45 +08:00
7Sageer	d88b3775c9	fix(agent-core-v2): count validation-rejected tool calls toward the repeat breaker (#2317) ... * fix(agent-core-v2): count validation-rejected tool calls toward the repeat breaker * style(agent-core-v2): remove inline implementation comments	2026-07-28 19:32:46 +08:00
STAR-QUAKE	de0ba9d065	fix(agent-core): count validation-rejected tool calls toward the repeat breaker (#2313) ... * fix(agent-core): count validation-rejected tool calls toward the repeat breaker Args-rejected calls returned before prepareToolExecution, so the breaker never counted them and the model could re-issue the same invalid call until maxSteps. Register them in finalizeToolResult so reminders fire at 3/5/8 and the turn force-stops at 12. * fix(agent-core): key parse-failed repeats on raw argument text Malformed JSON arguments normalize to {} on parse failure, which keyed every malformed-but-different attempt identically and could force-stop a turn whose calls were evolving rather than identical. Register skipped calls on the raw arguments text when parsing failed. --------- Co-authored-by: fengchenchen <fengchenchen@moonshot.ai>	2026-07-28 18:39:26 +08:00
Haozhe	d03a4886fd	feat(server): remove the 50 MiB upload size cap and stream uploads to disk (#2312) ... - add writeStream/putStream to the storage and blob store interfaces so large values are written incrementally (tmp + fsync + rename) instead of being buffered in memory - FileService.save now streams the request body straight to the blob store and only counts bytes for FileMeta.size - drop the multipart fileSize limit and the file.too_large error from the /files upload path (41301 stays in the wire protocol for session export)	2026-07-28 17:01:45 +08:00
liruifengv	b0f43aea28	feat(oauth): return structured managed usage rows (#2300) ... * feat(oauth): return structured managed usage rows Stop formatting plan-usage labels and reset hints into English strings at the oauth layer. The parser still absorbs backend field drift, but now emits a stable structured row (name / window{duration,unit} / used / limit / resetAt) that kap-server passes through and clients localize themselves: - window: normalized from duration/timeUnit (minute/hour/day/week), whole-hour minute windows fold to hours, unnamed summaries are the weekly limit - resetAt: absolute ISO timestamp; relative reset_in/ttl seconds are converted at parse time - TUI /usage panel formats labels and reset hints locally * refactor(oauth): parse managed usage strictly to the current payload shape Drop the defensive drift tolerance (alternate reset-time spellings, reset_in/ttl seconds, remaining-derived used, title/scope names, top-level duration/timeUnit, fuzzy time-unit matching) and parse only what the platform actually sends: numeric strings, resetTime, nested detail/window records, TIME_UNIT_* enums. * fix(node-sdk): update managed usage smoke example and facade tests for structured rows	2026-07-28 15:03:25 +08:00
Petrichor	cdbd33c13c	fix(kosong): fail fast on quota-exhausted 429 instead of retrying (#1857) ... * fix(kosong): fail fast on quota-exhausted 429 instead of retrying A 429 caused by an exhausted account quota or insufficient balance (Moonshot error.type "exceeded_current_quota_error", OpenAI "insufficient_quota") can never succeed on retry, yet it was classified as APIProviderRateLimitError and silently retried for the whole budget (10 attempts, ~3 minutes of backoff) with no UI feedback — the session appeared frozen on every request. Introduce APIProviderQuotaExhaustedError, minted in normalizeAPIStatusError from the structured body error.type/error.code forwarded by convertOpenAIError, with billing-anchored message patterns as a fallback for gateways that flatten the body to text. The new class is excluded from isRetryableGenerateError (fail fast, even when a retry-after header is present) and from isProviderRateLimitError (no swarm requeue/suspend). toKimiErrorPayload and translateProviderError map it to provider.api_error (retryable: false) instead of provider.rate_limit, and classifyApiError reports it as quota_exhausted in telemetry. agent-core-v2 mirrors the same fix. Transient rate-limit 429s keep the existing retry, backoff, and Retry-After behavior (verified end-to-end against a mock provider: quota body fails after attempt 1/10; rate-limit body still walks the full 10-attempt ladder). Behavior changes to note: quota-failed swarm subagents now fail instead of suspending indefinitely as "Rate limited...", and quota errors cross the wire as provider.api_error rather than provider.rate_limit. * fix(kosong): classify quota exhaustion in OpenAI Responses stream errors Responses response.failed / error SSE events carry no HTTP status and were minted by errorFromOpenAIResponsesEvent as either a rate-limit error (rate_limit_exceeded / embedded status_code=429) or a base ChatProviderError — and the base class falls into the retryable unclassified-failure fallback, so an insufficient_quota event still burned the whole retry budget on the openai_responses path. Route the event code and message through the same quota-exhausted check before the rate-limit branch, in kosong and the agent-core-v2 mirror. Covers all three entry paths (error events, response.failed, nested gateway frames) since they share the single converter. * style(agent-core-v2): drop inline comments per AGENTS.md header-only rule agent-core-v2 comments live solely in the top-of-file block, never beside functions or statements; the kosong twins keep the full rationale. * refactor(kosong,agent-core-v2): move quota-429 checks to vendor hook Per review on #1857: the knowledge of how a backend signals quota exhaustion is vendor-specific and must not run for every OpenAI-compatible provider from the shared conversion layer. - Add a convertError hook: ProtocolTrait.convertError in agent-core-v2 (single-value, last-declarer-wins, bound by composeOpenAIChatHooks / composeAnthropicHooks / traitConvertError) and an equivalent optional hook parameter on convertOpenAIError / convertAnthropicError. Bases consult it with the raw failure (SDK error on HTTP paths, raw event on the Responses in-stream path) after the abort guard, before their own rules. - Declare Moonshot's quota signals (exceeded_current_quota_error, billing wordings) on the Kimi side: kimiOpenAITrait and kimiAnthropicTrait in v2, the KimiChatProvider and KimiFiles catch sites in kosong, all through the new classifyKimiQuotaError. - Drop the options parameter from normalizeAPIStatusError and the shared quota code/pattern tables: the contract layer keeps only the vendor-neutral APIProviderQuotaExhaustedError type and its retry / rate-limit / wire-mapping semantics. - The OpenAI bases keep recognizing only OpenAI's own documented insufficient_quota code (HTTP and Responses stream events) as protocol knowledge of that wire. Behavior: kimi and openai provider types classify exactly as before; an unregistered vendor speaking Moonshot billing wordings through a plain openai transport now stays a retryable rate limit by design. * fix(kosong,agent-core,agent-core-v2): wire kimi quota hook fully Follow-up to the second review round on #1857, all four findings: - Kimi-over-Anthropic (legacy engine): AnthropicOptions gains the same optional convertError hook as the OpenAI bases, threaded through AnthropicStreamedMessage and every catch site, and the provider manager's anthropic route now passes classifyKimiQuotaError for provider type kimi — a quota-exhausted 429 over this transport previously still burned the retry budget. classifyKimiQuotaError now also walks error -> .error -> .error.error for the code/type, since the Anthropic SDK keeps the full body on .error instead of hoisting. - v2 telemetry: ApiErrorKind gains 'quota_exhausted' and classifyApiError checks APIProviderQuotaExhaustedError before the generic 429 branch, matching the legacy engine's reporting. - Hook contract: converted ChatProviderErrors now pass through before the vendor hook is consulted in convertOpenAIError / convertAnthropicError (both engines), so the hook sees each raw failure exactly once even when a stream-minted error crosses an outer catch; tests assert the single consult. - protocolTrait: the convertError member doc shrinks to the concise style and the consult contract moves into the file header's composition rules. * test(kosong,agent-core,agent-core-v2): lock quota hook assembly paths Third review round on #1857: - Fix the v2 anthropic base header and AnthropicHooks doc still claiming withThinking is the only hook. - Drop the two remaining non-header JSDoc blocks in protocolTrait.ts per the AGENTS.md header-only rule; the consult contract already lives in the file header. - Update the ProtocolTrait contract test to the seventeen-hook shape (convertError included) and cover the traitConvertError binding. - Add real-assembly regression probes: the v2 registry composes a (kimi, anthropic) provider whose mocked SDK client throws a Moonshot quota 429 and generate rejects with the non-retryable APIProviderQuotaExhaustedError (a plain anthropic composition keeps the same 429 retryable); the legacy ProviderManager routing test asserts convertError is classifyKimiQuotaError on the kimi-anthropic route and absent for plain anthropic; the legacy provider threads options.convertError to its generate catch. * test(kosong,agent-core-v2): cover KimiFiles quota 429 and drop stale docs Fourth review round on #1857: - Drop the AnthropicHooks member JSDoc (its content already lives in the anthropic.ts and anthropicHooks.ts file headers) and fix the anthropic contrib header still calling the hook set single-hook. - Add the missing KimiFiles regression in both engines: a mocked files client rejecting with a Moonshot quota 429 makes uploadVideo reject with the non-retryable APIProviderQuotaExhaustedError, locking the classifyKimiQuotaError argument at the upload catch sites.	2026-07-28 14:35:12 +08:00
Haozhe	e556088458	feat(tree-sitter-bash): add a pure-TypeScript bash parser and an agent-core-v2 bashParser service (#2016) ... * feat(tree-sitter-bash): scaffold pure-TypeScript bash parser package Add packages/tree-sitter-bash with the SyntaxNode tree model (UTF-16 code-unit offsets, tree-sitter-bash named-node type names), a parse budget (deadline + node cap) with an aborted ParseResult variant, and a placeholder parse() to be replaced by the real lexer/parser. * feat(tree-sitter-bash): add lexer and core recursive-descent parser Cover the permission-analysis grammar subset: lists, pipelines, commands, words, quotes, expansions, command/process substitution, subshells, the full redirect operator set, heredocs and comments, with tree-sitter-bash named-node type names. Long scan loops check the parse deadline via budget.progress() so large literals cannot exhaust the node cap; parse depth is bounded on all recursion chains. * feat(tree-sitter-bash): support the full bash grammar Add compound commands (if/while/until/for/c-style-for/select/case/ function/compound/do_group), test commands with reference-exact extglob/regex right-hand-side rules, a Pratt expression engine for arithmetic expansion shared across arith/c-for/test modes, arrays and subscripts, declaration/unset commands, ansi-c/translated strings and brace expressions. Reserved words are recognized only in statement position. Case-aware balanced scanning keeps command substitutions intact around case items, expression leftovers are kept as ERROR nodes instead of dropped, and recursion depth is bounded per chain (substitution 150, parse 500, lexer scan 1024). * test(tree-sitter-bash): add differential fixtures, corpus, fuzz and perf suites Turn the ad-hoc wasm comparison work into permanent infrastructure: a differential helper pinning reference-equivalent and known-difference samples against the real tree-sitter-bash wasm (478 match / 79 known-diff fixtures plus the official v0.25.0 corpus), a three-way consistency check between fixtures, the known-difference registry and the README, deterministic seeded fuzz with tree-integrity assertions, and performance smoke tests. Converges 15 further divergence groups found by systematic probing and documents the rest; the full suite is 853 tests in ~4s. * feat(agent-core-v2): add App-scope bashParser service Wrap the pure @moonshot-ai/tree-sitter-bash package as IBashParserService (L1, no dependencies): parse(source, options) returns a wire-safe BashParseResult whose nodes drop the cyclic parent link so trees can cross the RPC boundary. Budget exhaustion surfaces as { ok: false, reason: 'aborted' }, malformed input as hasError — never a throw. * chore: add changeset for the bash parser service * chore: update pnpmDeps hash after adding tree-sitter-bash package * fix(agent-core-v2): register bashParser service with ScopeActivation The registration was written against the removed InstantiationType API (#/_base/di/extensions); switch to ScopeActivation.OnDemand from #/_base/di/scope so the package typechecks and the service registers. * feat(kimi-inspect): add Bash Parser view Add a fourth icon-rail tab that exercises the App-scope bash parser service over the debug RPC surface: a source textarea with a parse budget (timeoutMs / maxNodes), a dropdown of curated examples adapted from the tree-sitter-bash differential fixtures, and an expandable syntax tree with per-node type, UTF-16 range and leaf text, plus hasError / aborted / node-count badges. * fix(agent-core-v2): snapshot bash syntax trees iteratively A long left-associative chain (e.g. an arithmetic expression with a few thousand operands) parses into a tree thousands of levels deep while still within budget; the recursive DTO conversion then overflowed the JS call stack and made parse throw RangeError, breaking the never-throws contract. Convert the tree with an explicit stack, the same approach as the parser's own materialize. * feat(kimi-inspect): add a deep-arithmetic example to the Bash Parser view A thousand-operand left-associative chain fills the textarea with a thousand-level binary_expression tree — the shape that once overflowed the DTO conversion. Deeper chains still parse in-process but cannot cross the JSON RPC transport (V8 call-stack limit in serialization), so the example stays within the wire limit. * chore: update pnpmDeps hash for the rebased lockfile The rebase onto main merged pnpm-lock.yaml, invalidating the recorded fetchPnpmDeps hash; use the hash CI computed for the merged lockfile.	2026-07-28 14:29:08 +08:00
Haozhe	7e30add445	feat(kap-server): add global fs:mkdir endpoint (#2281) ... * feat(kap-server): add global fs:mkdir endpoint Add POST /api/v1/fs:mkdir to create a directory on the host filesystem by absolute path, backing the folder picker's "new folder" action. Implemented directly on node:fs/promises.mkdir in the transport layer for now, non-recursive by design, with wire errors mapped to the existing fs.* codes (40001/40409/40411/40919). * test(kap-server): update api surface snapshot for fs:mkdir * test(kap-server): stop export tests from holding server.close() open The export download tests reused pooled undici keep-alive connections, so afterEach's server.close() could wait out fastify's 72s default keepAliveTimeout and die on the 10s hook timeout (flaky on CI). Send connection: close on the streamed export requests, matching the fs:content tests.	2026-07-28 10:50:39 +08:00
7Sageer	77618e38c3	feat(kap-server): wire cloud telemetry for engine events (#2230) ... * feat(kap-server): wire cloud telemetry for engine events The v2 engine registers a full telemetry event catalog, but kap-server never attached an appender, so events from web-hosted sessions were dropped to the null appender. Add an opt-in `telemetry` start option that attaches a CloudAppender (app_name kimi-code-cli, ui_mode web, matching the v1 `kimi web` host conventions), still gated by the config `telemetry` toggle, with periodic flush and a bounded flush on close. The option defaults off so tests never post to the real endpoint; the CLI's `kimi web` host enables it. * fix(kap-server): honor telemetry disable environment * fix(agent-core-v2): isolate session telemetry context * fix(kap-server): seed telemetry client version * fix(cli): share telemetry shutdown deadline * fix(telemetry): make shutdown ownership durable * fix(kap-server): keep telemetry shutdown best-effort	2026-07-27 17:18:16 +08:00
Haozhe	a77ee03829	feat(agent-core-v2): add hostIdentity domain for host-overridable system prompt identity (#2144) ... * feat(agent-core-v2): add hostIdentity domain for host-overridable system prompt identity - add app-scope hostIdentity domain (L3) with productName / replyStyleGuide overrides and hostIdentitySeed for composition roots - render ${product_name} and ${reply_style_guide} in the base system prompt, falling back to CLI defaults when the host provides no override - seed the variables from AgentProfileService via IHostIdentity - expose hostIdentity on kap-server's ServerStartOptions * chore: add changeset for host identity system prompt * fix(agent-core-v2): adapt hostIdentity registration to ScopeActivation The DI refactor on main removed _base/di/extensions (InstantiationType); register with ScopeActivation.OnScopeCreated instead, and regenerate the state manifest for the new SystemPromptContext fields. --------- Co-authored-by: liruifengv <liruifeng1024@gmail.com>	2026-07-27 16:55:41 +08:00
7Sageer	3b017821cf	feat(kap-server): accept secondary_model in the config API (#2228) ... * feat(kap-server): accept secondary_model in the config API POST /api/v1/config now accepts secondary_model, persisted to the [secondary_model] config section via the generic per-domain dispatch. GET /config also hides the synthesized __secondary__ derived entry from the models view, matching the GET /models listing. * Update config-api-secondary-model.md Signed-off-by: 7Sageer <sag77r@hotmail.com> --------- Signed-off-by: 7Sageer <sag77r@hotmail.com>	2026-07-27 16:45:18 +08:00
liruifengv	48bf3d4c28	feat(kap-server): bundle the desktop app log into session exports on request (#2223) ... * feat(kap-server): bundle the desktop app log into session exports on request * refactor(agent-core-v2): align the desktop log export with repo conventions	2026-07-27 12:07:58 +08:00
7Sageer	d40d0d305d	refactor(agent-core-v2): make undo domain-owned (#2055) ... * refactor(agent-core-v2): rebuild undo as wire-level journal rewind Replace the compensating context.undo op with a wire-layer rewind primitive: a log.cut control record with a persisted target, applied uniformly by the wire during fold. Turn boundaries become first-class (TurnIndexModel indexing turn.prompt record positions), models declare a temporal classification (rewindable), and a single IAgentRewindService owns the undo pipeline (quiesce -> precheck -> cut -> reconcile) with all entry points converged. - wire: log.cut record, rewindable model flag, re-fold rebuild; OpApplyContext.recordIndex for position-aware reducers - rewind service: aborts the active turn, cancels in-flight compaction, preserves the pending queue, rebases measured tokens, reconciles lastPrompt, tracks conversation_undo - todo list, plan mode, task-notification delivery and the turn index now rewind together with the undone turns - transcript reducer applies cut ranges so snapshot/messages surfaces stay consistent with the model context - REST/RPC/debug undo entry points converge on the rewind service; TUI parses the v2 undo-unavailable error shape - legacy context.undo records keep replaying for old journals * refactor(agent-core-v2): keep undo domain-owned * refactor: enhance /undo functionality for consistency and safety, including todo list rollback and improved event handling * chore: clean up undo changeset artifacts * refactor: rebuild rewind consistency * fix: make conversation undo durable and consistent * fix(agent-core-v2): stabilize undo restoration * fix: keep TUI undo on legacy error contract * refactor(agent-core-v2): drop unused full compaction cancel API Undo now rejects with session.busy while compaction runs instead of cancelling it, so the awaitable cancel() added for the earlier rewind semantics has no callers left. Remove it from the interface and implementation; the RPC cancel path keeps using the task abort controller directly. * fix(agent-core-v2): remove injected context on undo * chore(agent-core-v2): regenerate wire manifest * docs(agent-core-dev): rename rewind to undo in layer table * fix(agent-core-v2): undo prompt-owned image reminders * refactor: remove transcript undo reconciliation * fix(kap-server): map undo busy errors * refactor(agent-core-v2): rename undo participant registry and attribute checkpoint depth - Rename IAgentConversationUndoReconciliationRegistry to IAgentConversationUndoParticipantRegistry (conversationUndoParticipants). - Return the limiting model from checkpointDepth and include it in the SESSION_UNDO_UNAVAILABLE details; report checkpoint_lost instead of compaction_boundary when no compaction explains the missing depth. - Add a registry invariant test: every model reacting to context.* ops must be registered via defineCheckpointedModel or explicitly exempt. * Delete .changeset/fix-undo-injections.md Signed-off-by: 7Sageer <sag77r@hotmail.com> --------- Signed-off-by: 7Sageer <sag77r@hotmail.com>	2026-07-27 11:09:26 +08:00
Haozhe	cc9b25e132	fix(kap-server): flush public and control WS frames immediately (#2221) ... Only subscription events now enter the timed coalescing buffer. server_hello, acks, resync_required, and global session events join the outbound FIFO and flush it right away, so they no longer wait behind the flush window and cannot overtake earlier buffered subscription frames. The broadcaster marks global fan-out with the new BroadcastDelivery 'immediate' lane.	2026-07-27 10:58:38 +08:00
Luyu Cheng	0cef160c4b	fix(agent-core): continue goal pursuit when a goal turn hits the per-turn step limit (#2210) ... * fix(agent-core): continue goal pursuit when a goal turn hits the per-turn step limit * chore(agent-core-v2): regenerate state manifest * fix(agent-core): start goal pursuit when the goal-creating turn hits the step limit * refactor(agent-core-v2): drop step-cap narration from goal module and test	2026-07-27 00:30:33 +08:00
Haozhe	bf8e967d5c	refactor(agent-core-v2): register tools as scoped services (#2196) ... * refactor(agent-core-v2): register agent tools as DI services with profile-aware activation - replace module-level registerTool + Eager AgentBuiltinToolsRegistrar with registerAgentTool double registration (Agent-scope DI service + contribution table) - add toolActivation domain: AgentToolActivationService filters contributions by the bound Profile's tool policy using declared names, resolves instances lazily via accessor.get, and re-activates on agent.status.updated - rename BuiltinTool to AgentTool service interface; tools become Agent-scope services with decorator-injected dependencies (e.g. AgentTool -> SubagentTool/ISubagentTool) - AgentLifecycleService.create runs one activation pass after restore and profile binding so tools reflect the Profile before the first turn - update tool registrations, scripts, and tests; add toolActivationService tests * refactor(agent-core-v2): centralize builtin tools under agent/tools - move builtin tools from scattered domain folders (plan/tools, goal/tools, os/backends/node-local/tools, task/tools, etc.) into a unified agent/tools/ directory - split each tool into a kebab-case definition file (bash.ts) and a registration file (bashTool.ts) pairing with its prompt markdown - update imports across src, tests, kap-server, and TUI comments * fix(agent-core-v2): keep agent tools out of scope-creation instantiation Main's instantiateAll constructs every registered service at scope creation, but agent tool constructors may legitimately throw when their host capability is absent (e.g. WebSearchTool without a configured provider), and profile-aware activation must stay the only resolution path so the runtime registry holds real instances, never proxies. - add SyncDescriptor.instantiateWithScope (default true) and let registerScopedService opt registrations out of the instantiateAll sweep - registerAgentTool passes the opt-out, restoring lazy activation-driven construction on top of the eager-scope semantics - regenerate docs/state-manifest.d.ts * refactor(agent-core-v2): replace delayed DI with scope activation - add explicit OnScopeCreated and OnDemand activation modes - remove delayed proxy and idle initialization support - migrate service registrations, tests, and DI guidance	2026-07-26 14:32:59 +08:00
Haozhe	7799bd7346	feat(agent-core-v2): add per-scope keyed state container (#2192) ... * refactor(agent-core-v2): instantiate all registered services eagerly at scope creation - add instantiateAll to scope creation: every service registered for a scope tier is constructed when the scope is created, following the static dependency graph; a failing constructor fails scope creation - drop the hand-maintained eager-resolution lists: igniteEagerServices in AgentLifecycleService, the force-instantiated session services in SessionLifecycleService, and the kosong config bridge get in bootstrap - update DI docs and agent-core-dev skill guidance for the new semantics - adjust affected tests (registry hygiene, timer draining, listener ordering) and add scope-tree coverage for eager instantiation * feat(agent-core-v2): add per-scope keyed state container (state domain) - add _base StateRegistry: typed StateKey/defineState descriptors with register/get/set, per-key onDidChange and global onDidChangeAny events, and BugIndicatingError on duplicate or unregistered key access - bind thin per-scope services at each tier: IStateService (App), ISessionStateService (Session), IAgentStateService (Agent), so scoped plain-data state lives in one observable container that dies with the scope - register the state domain in the layer check script and export the new services from the package index - add StateRegistry unit tests and scoped resolution tests * refactor(agent-core-v2): move session-scope service state into ISessionStateService - add StateRegistry.snapshot() with JSON-safe serialization for Map/Set/Date/circular values - migrate plain-data fields of 13 session-scope services (cron, interaction, sessionActivity, agentProfileCatalog, fs, fsWatch, log, metadata, skillCatalog, toolPolicy, workspaceCommand, workspaceContext) to defineState keys registered in sessionState - register SessionStateService in affected tests and add test/state/stubs.ts helper * feat(kimi-inspect): rework chat layout with session pane and tabbed right dock - add SessionPane column next to the sidebar: Services tab (pending interactions + session Service panels) and State tab polling ISessionStateService.snapshot() every second, rendered as a live diff tree - merge the transcript audit panel and the agent inspector into one RightPanel with Audit/Agent tabs that keep panel state across switches - extract InteractionsCard from Inspector into its own component - collapse multiline strings in StateTree into a compact hover-preview button with a viewport-clamped fixed popup - test: cover StateRegistry.snapshot() conversions (Map/Set/circular); pass a session state service to SessionInteractionService in kap-server test fakes - update the AGENTS.md project map for the new layout * refactor(agent-core-v2): move agent-scope service mutable state into agentState Register each Agent-scope service's mutable fields into the agent-state container (IAgentStateService) via defineState keys, and access them through get/set accessors backed by states.get/set. Promise locks, disposables, and other mechanism-only fields stay plain instance fields. - define per-service state keys (e.g. activityView.*, goal.*, toolDedupe.*) and register them in each service constructor - replace direct field reads/writes with states-backed accessors across the touched services - update the affected unit tests for the new IAgentStateService dependency * fix(agent-core-v2): collapse class instances in state snapshots - StateRegistry.snapshot() now stops at custom-prototype boundaries and emits '(ClassName)' markers, so resource graphs reachable from registered values (e.g. tool instances holding service references) can no longer exhaust the heap during export; plain data keeps recursing - add agent-scope state test covering the full assembled agent scope - kimi-inspect: extract the session State card into a shared StateCard and add an agent State tab in RightPanel polling IAgentStateService.snapshot() - document the per-scope state container pattern in the agent-core-dev skill * refactor(agent-core-v2): keep resource-holding state out of the state container - move fields holding live resources (tool entries, managed tasks, turn jobs, prompt records, MCP registrations, profile callbacks) back to plain private fields so the agent state service only holds snapshot-safe plain data - add gen:state-manifest script that statically collects defineState keys and their register call sites into docs/state-manifest.d.ts - add state manifest freshness test and update agent-state docs * chore: add changeset for session-scope state container * docs(agent-core-v2): move eager-instantiation notes into the scope.ts header * fix(kimi-inspect): suppress stale state tree while switching state owner	2026-07-26 10:10:15 +08:00
Luyu Cheng	c497af60e6	fix(tui): steer user messages into the running turn while a goal is active (#2153) ... * fix(tui): steer user messages into the running turn while a goal is active * fix(tui): reset request state when steering mid-goal input fails * fix(agent-core): update session prompt metadata on steer	2026-07-25 02:12:24 +08:00
Liu Zhongnuo	f06eb5c60e	feat(agent-core): defer registered user tools (#2119) ... * feat(agent-core): defer registered user tools Allow hosts to mark registered user tools as deferred so select_tools can discover and load their schemas on demand in both agent-core implementations. * fix(agent-core): hide unregistered deferred tools Filter loaded deferred user-tool schemas from provider history and loaded-state checks after unregister while preserving canonical history and disconnected MCP behavior. * fix(agent-core-v2): drop stale inline schemas Stop treating previously loaded user-tool schemas as active after the same tool is re-registered for inline disclosure.	2026-07-24 17:48:36 +08:00
Haozhe	3615b5da9f	refactor(agent-core-v2): drop definition-level capability resolution (#2142) ... Capability resolution now falls back from trait hooks (last declarer wins) to the protocol base catalog to UNKNOWN_CAPABILITY; the definition layer is removed from the chain. - remove ProviderDefinition.capability and the definition branch in ProtocolAdapterRegistry.explainCapability - kimi contrib drops its UNKNOWN_CAPABILITY declarations, so base-known model ids resolve through the base catalog instead of being suppressed by a vendor-level UNKNOWN - update composition tests for the new resolution order	2026-07-24 16:04:02 +08:00
liruifengv	a2401cc1ed	feat(kap-server): add provider write endpoints and models.dev/registry import (#2110) ... * fix(agent-core-v2): honor explicit undefined field deletes in modelsToToml The models TOML transform merged each entry as { ...oldRaw, ...converted }, so a field the new record carried with an explicit undefined — deleted from 'converted' by setDefined — was put right back from the old on-disk raw. Field-level clears issued through config.replace (e.g. the provider write routes dropping base_url, display_name, capabilities) only took effect in memory and resurrected on the next boot. Merge via setDefined onto the old raw directly, matching providerEntryToToml. * feat(kap-server): add provider write endpoints and models.dev/registry import Add the provider management write surface plus server-proxied catalogs so API clients (the desktop app first) can manage providers without editing config.toml by hand: - POST/PUT/DELETE /providers: manual create, replace-style edit (new_id rename with pointer migration, tri-state api_key, form-unknown fields merged through), and delete with real alias removal. - GET /providers/{id} additionally reveals the stored api_key for edit prefill on the loopback+bearer transport; the list stays redacted. - GET /catalog/providers[{id}]: pruned models.dev directory with import eligibility resolved server-side (10-min cache, stale fallback, built-in snapshot, shared in-flight fetch). - POST /providers:import_catalog and :import_registry (collection actions next to :refresh): catalog/registry imports with TUI-aligned remove-then-apply refresh semantics, OAuth-managed guards, and the registry source blob that scheduled refreshes rediscover. Write-path consistency: field-level clears assign explicit undefined (the TOML transforms only drop keys that way), import refreshes swap aliases in two passes so stale on-disk fields cannot ride along, re-imports keep the stored credential when api_key is omitted, and multi-step writes serialize through a process-local chain. Global default_provider/default_model are never modified by provider writes. * fix(kap-server): cache the built-in catalog fallback and guard alias-key collisions on replace - Cache the built-in snapshot on upstream failure, so an offline install no longer pays the full upstream timeout on every catalog call. - Reject a PUT whose rebuilt alias keys collide with an alias owned by another provider (foreign-prefix keys are global), checked before any write so a collision never lands half the edit. * fix(kap-server): make toCatalogProviderItem's switch explicit for older TS control-flow * chore(agent-core-v2): drop inline rationale per package comment conventions * feat(kap-server): seed default_model on provider create/import when none is configured A fresh setup has no global default model, so the first provider added through the write endpoints left GET /auth permanently unready and new sessions without a selected model. Seed default_model from the created provider's default (or first) model on POST /providers, :import_catalog and :import_registry when — and only when — no default is configured; an existing pointer is never moved, not even a dangling one. * refactor(agent-core-v2): own the models.dev provider import in the engine kap-server's provider write endpoints imported @moonshot-ai/kosong and @moonshot-ai/kimi-code-oauth directly for the models.dev browse/import and custom-registry import. Move that capability behind a new App-scope IModelsDevImportService so the server only talks to agent-core-v2: - app/kosongConfig/modelsDev.ts mirrors the third-party api.json schema and hosts the pure translation (wire inference, endpoint resolution, model pruning, thinking/gateway overrides) ported from kosong's catalog.ts; kosong's own type surface stays limited to built-in vocabulary - modelsDevUpstream.ts owns the directory fetch, 10-minute cache, and built-in snapshot fallback; modelsDevImportService.ts owns the import orchestration (tri-state api_key, two-pass alias swap, OAuth-managed rejection, serialized writes) with coded modelsDev.* errors - kap-server routes shrink to wire mapping (zod schemas + numeric error code mapping) and drop the kosong/oauth dependencies The /api/v1 surface is byte-identical: the apiSurface snapshot and all existing modelCatalog tests pass unchanged. * chore: drop the branch's changesets * fix(agent-core-v2): route the models.dev always-thinking exemption through kosong The zero-tolerance vendor-name gate forbids a 'kimi' compare outside the kosong layer, and the engine-side models.dev port carried one inline. Kosong (the vendor layer, owner of thinking semantics) now answers the verdict — wireHasProtocolThinkingDisable — and kosongConfig/modelsDev.ts asks it instead of comparing wire ids. --------- Co-authored-by: haozhe.yang <yanghaozhe@moonshot.ai>	2026-07-24 14:00:16 +08:00
7Sageer	7b62ed5b2c	feat: support a configurable secondary model for subagents (#2064) ... * feat: support a configurable secondary model for subagents * refactor: move the subagent model config to a consumer-neutral [secondary_model] The secondary model becomes a model-domain concept next to default_model so future consumers beyond subagents can share it: [secondary_model] model / effort in config.toml, KIMI_SECONDARY_MODEL / KIMI_SECONDARY_EFFORT env overrides, and the Agent / AgentSwarm per-spawn choice renamed from "subagent" to "secondary". * docs: replace "v2 engine only" notes with the concrete effective surfaces State that [secondary_model], its env overrides, the Agent/AgentSwarm model parameter, and SYSTEM.md take effect only under kimi web and experimental kimi -p (the TUI ignores them), and that --agent / --agent-file are available only under experimental kimi -p. Also drop the SYSTEM.md claim of parity with --agent/--agent-file, which was inaccurate: SYSTEM.md is an agent-core-v2 app-domain feature and also applies under kimi web, while the flags are gated at the CLI. * fix: review follow-ups for the subagent secondary model - Drop the "cheaper" claim from the Agent/AgentSwarm model parameter descriptions and the advertised model list — the secondary model is not necessarily the cheaper one. - Downgrade the changeset to patch, note the kimi web / experimental kimi -p effective surface, and tighten the wording. - Remove the onWillRestore stub fields from two lifecycle stubs; they belong to upcoming lifecycle work, not to this change. * fix(agent-core-v2): prevent ghost agents from invalid model bindings * fix: narrow the secondary-model error hint to missing-alias failures The model catalog's not-configured throw now carries details.model, and wrapSubagentModelError only decorates errors whose details.model matches the bound model. Malformed [models.*] entries and unrelated config.invalid failures during agent creation pass through untouched instead of being misattributed to an invalid secondary-model alias. * fix: mark subagent resume semantics as breaking * chore(agent-core-v2): follow header-only comment convention * fix(agent-core-v2): await agent restore preparation * feat(agent-core-v2): support agent model preferences * Update subagent-secondary-model.md Signed-off-by: 7Sageer <12210216@mail.sustech.edu.cn> * fix: validate secondary models before agent creation * Delete .changeset/secondary-model-startup-warning.md Signed-off-by: 7Sageer <12210216@mail.sustech.edu.cn> * Add secondary_model config section for subagents Individual agents can override this via the new `model_preference` field in their agent file. Signed-off-by: 7Sageer <sag77r@hotmail.com> * feat(agent-core-v2): support override patches in [secondary_model] The recipe is now `model` plus the flattened ModelOverride field set. With any patch field set, a config overlay synthesizes a derived registry entry (base copy, patch merged into overrides, aliases dropped) so subagent spawning rides the standard effectiveModelConfig merge; with none, subagents bind the pointed entry directly. `default_effort` replaces `effort` (KIMI_SECONDARY_EFFORT rebinds) and doubles as the explicit subagent thinking; unset, thinking resolves naturally instead of inheriting the caller. The overlay strips the derived entry (and any defaultModel pointer to it) from writes, and the kap-server GET /models route hides it from pickers. * fix(agent-core-v2): fire section events for overlay-rewritten domains rebuildEffective only committed the caller-named domains, so a ConfigEffectiveOverlay or section env binding that rewrote a sibling domain (setting [secondary_model] synthesizes a derived models entry; removing the recipe retracts it) left consumers of the models section stale. Widen the commit candidates with every domain the recompute actually changed; commit() deepEqual-guards each candidate, so the widening costs nothing. * feat(agent-core-v2): gate secondary model behind experimental flag * Update subagent-secondary-model.md Signed-off-by: 7Sageer <sag77r@hotmail.com> --------- Signed-off-by: 7Sageer <12210216@mail.sustech.edu.cn> Signed-off-by: 7Sageer <sag77r@hotmail.com>	2026-07-24 12:16:53 +08:00
Kai	66f611aae9	fix: echo thinking under the reasoning field the endpoint actually uses (#2104) ... OpenAI-compatible endpoints disagree on the wire field for reasoning content: `reasoning_content` (DeepSeek/Moonshot convention, older vLLM) vs `reasoning` (OpenAI GPT-OSS guidance, current vLLM — which also only accepts `reasoning` on the request side, vllm-project/vllm#38488). Reading only `reasoning_content` silently dropped thinking, and the lost thinking then never made it back into later requests. Add per-endpoint dialect detection: inbound responses are scanned in priority order (reasoning_content, reasoning_details, reasoning; first string value wins), the carrying key is remembered, and outbound messages echo thinking under the same key (default reasoning_content). An explicit `reasoningKey` / `reasoning_key` config still pins the dialect. - kosong: shared reasoning-key module; wire the kimi and openai-legacy providers; the dialect cell is shared across per-step provider clones. - agent-core-v2: same mechanism in the vendored openai-legacy base; the kimi trait no longer pins reasoning_content (the default already is), keeping Moonshot wire behavior byte-identical while adapting to vLLM. - agent-core: memoize the base provider behind ConfigState.provider so the detected dialect survives across turns instead of being rebuilt per access.	2026-07-23 21:15:00 +08:00
Haozhe	d751b6796c	feat(kap-server): global session work status, transcript subscribe_v2, and plan endpoint (#2094) ... * feat(session): add core work aggregate and consume it at WS/REST edges - agent-core-v2: add the sessionActivity domain (ISessionActivityView, Session scope) folding each agent's activity view and the interaction kernel into busy / main_turn_active / pending_interaction / last_turn_reason, firing cause-tagged change events only on real tuple changes - kap-server: resolveSessionFacts reads the core view; the WS broadcaster drops its per-agent fold/dedup and delegates to the view, deferring turn_ended emissions until after the turn.ended frame so busy:false never precedes it; global events now fan out to every established connection via addGlobalTarget without any subscription - kimi-inspect: add src/activity (GlobalEventsWs + SessionActivityHub + useSessionActivities) consuming the global push; the Sidebar renders running / approval / question / failed badges per session * feat(kap-server): decouple transcript stream from agent_filter - transcript frames (transcript.reset/ops) are governed by the per-agent transcript grades alone and bypass the legacy agent allowlist; the filter still gates session_event delivery - client_hello is handshake-only in code: hello and subscribe share one attach path, and hello's inline subscription fields are deprecated (subscriptions made optional) in both protocol packages - flush deferred work_changed(busy:false) from a microtask so it always lands after the matching turn.ended frame * fix(kimi-inspect): restore session activity hub in the browser - bind the default fetch in SessionActivityHub: a member call hits the browser's Illegal invocation (receiver is not the global object), and the swallowed error left the REST seed never firing, so the Sidebar badges never populated on refresh - create the hub in useEffect + useState instead of useMemo: under StrictMode the first mount's cleanup closes the memo-created hub for the rest of the page's life * chore: add changeset for the global session work status push * feat(kap-server): add transcript plan endpoint for ExitPlanMode calls - add GET /sessions/{id}/transcript/plan projecting one ExitPlanMode call's plan info (content, path, options, review outcome) from the first available fact: the linked approval interaction's request display, the live tool frame's display, or the tool result output - add TOOL_CALL_NOT_FOUND (40416) error code - add transcriptPlanResponseSchema to the transcript contract - cover live, auto-mode, cold-rebuild, and error paths in tests - update the API surface snapshot and AGENTS.md * feat: move transcript subscriptions to subscribe_v2 and extend the plan endpoint - add subscribe_v2 / unsubscribe_v2 WS control frames as the only carrier of per-agent transcript grades and the transcript_since cursor; client_hello / subscribe no longer accept transcript fields - serialize control-frame handling per connection so interleaved attaches cannot overwrite fresher subscription state - make tool_call_id optional on GET /sessions/{id}/transcript/plan: omitted lists every ExitPlanMode call with recoverable plan content as a plans array - add a Plan lookup card to the kimi-inspect agent inspector via fetchTranscriptPlan - add TOOL_CALL_NOT_FOUND (40416) to the shared protocol error codes - update AGENTS.md and add changesets	2026-07-23 18:48:25 +08:00
7Sageer	5fdbdb4a22	feat: configure web search/fetch services via KIMI_WEB_* env vars (#2096) ... * feat: configure web search/fetch services via KIMI_WEB_* env vars KIMI_WEB_SEARCH_BASE_URL / KIMI_WEB_SEARCH_API_KEY and KIMI_WEB_FETCH_BASE_URL / KIMI_WEB_FETCH_API_KEY overlay the [services] config section field by field in both engines (env wins over config.toml), so the WebSearch and FetchURL backends can be pointed at a Moonshot service without OAuth login. The v2 engine now also honors the explicit [services.moonshot_fetch] section (config > managed OAuth > local), which it previously parsed but never consumed. * fix(agent-core-v2): guard stripServicesEnv against clearing the services section config.replace(SERVICES_SECTION, undefined) — the logout deprovisioning path in authService — passes undefined straight into the section strip, and the composed stripServicesEnv dereferenced it (value[key]), throwing TypeError and aborting logout mid-cleanup. Add the same isPlainObject guard the other strip implementations (stripProvidersEnv, stripEnvBoundFields) already have, plus a regression test that also locks in the env overlay staying effective after the file value is cleared. * style(agent-core-v2): follow header-only comment convention * fix: isolate env web service credentials * Add env vars for web search and fetch services The new environment variables `KIMI_WEB_SEARCH_BASE_URL`, `KIMI_WEB_SEARCH_API_KEY`, `KIMI_WEB_FETCH_BASE_URL`, and `KIMI_WEB_FETCH_API_KEY` take priority over the corresponding fields in `config.toml`. The `kimi web` backend now honors the `[services.moonshot_fetch]` config section. Signed-off-by: 7Sageer <sag77r@hotmail.com> --------- Signed-off-by: 7Sageer <sag77r@hotmail.com>	2026-07-23 17:36:16 +08:00
7Sageer	527d485d92	feat: add global default MCP server timeout configs (#2065) ... * feat: add global default MCP server startup timeout config Add a `[mcp] startup_timeout_ms` config.toml section with a `KIMI_MCP_STARTUP_TIMEOUT_MS` env override as the global default MCP server connection (startup + tool discovery) timeout. Precedence: per-server `startupTimeoutMs` in mcp.json > env var > config.toml > built-in 30s default. * feat: add global default MCP tool call timeout config Extend the `[mcp]` section with `tool_timeout_ms` and the `KIMI_MCP_TOOL_TIMEOUT_MS` env override as the global default for single MCP tool calls, mirroring the startup timeout: a per-server `toolTimeoutMs` in mcp.json still wins, and unset entries fall back to the SDK built-in 60s default. * chore: shorten the mcp timeouts changeset * feat(agent-core): add global default MCP server timeout configs Port the `[mcp]` section (`startup_timeout_ms` / `tool_timeout_ms`) and the `KIMI_MCP_STARTUP_TIMEOUT_MS` / `KIMI_MCP_TOOL_TIMEOUT_MS` env overrides to agent-core (v1), mirroring the v2 semantics: per-server fields in mcp.json > env vars > config.toml > built-in defaults. The v1 TOML loader gains explicit `mcp` read/write mappings, and both connection-manager construction sites (Session, testGlobalMcpServer RPC) pass the resolved defaults through. * fix(agent-core): validate and apply MCP timeout defaults * fix: propagate MCP startup timeout to SDK requests * refactor(agent-core-v2): resolve MCP default timeouts at connect time Keep the session connection manager synchronously lazy instead of gating its existence on config readiness: resolveDefaultTimeouts is read from the mcp config section at each (re)connect, so AgentMcpService's eager construction stays unconditionally safe and reconnects pick up changed preferences. The initial connect still awaits config.ready for a deterministic snapshot. Also restore the ISessionMcpService method docs, bump the changeset to minor, and fix the v1 env-parse comment. * chore(agent-core-v2): regenerate config manifest for the mcp section * Add global default MCP server timeouts configuration Specify the new global default MCP server timeouts in both the config file and environment variables. Signed-off-by: 7Sageer <sag77r@hotmail.com> --------- Signed-off-by: 7Sageer <sag77r@hotmail.com>	2026-07-23 12:41:08 +08:00
Haozhe	5240b5c83c	feat(agent-core-v2): add generated config and wire-protocol manifests (#2086) ... * feat(agent-core-v2): add generated config section manifest - add scripts/gen-config-manifest.mts: drains the live registerConfigSection / registerConfigOverlay contributions and renders docs/config-manifest.toml in the on-disk config.toml shape (owner, scope, registered defaults, env bindings, schema fields) - add a gen:config-manifest package script (--check mode included) and a freshness test that rebuilds the manifest and compares byte-for-byte - point the agent-core-dev config skill and the package AGENTS.md at the generated manifest instead of the stale hand-maintained ownership map * feat(agent-core-v2): add generated wire-protocol manifest - add scripts/gen-wire-manifest.mts to generate docs/wire-manifest.d.ts from defineOp registrations (payload interfaces, persist policy, toEvent, cross-reducers) plus a WirePayloadMap - extract shared JSON Schema helpers from gen-config-manifest.mts into scripts/lib/jsonSchema.mts - add gen:wire-manifest script and wireManifest.test.ts freshness check - document the manifest in packages/agent-core-v2/AGENTS.md * fix(agent-core-v2): keep array-of-tables manifest sections fully commented A bare `[hooks]` header parses as a plain table, which array sections reject on load; emit only the commented `[[hooks]]` shape so the manifest matches the on-disk config.toml shape it documents. Addresses a Codex review comment on PR #2086.	2026-07-23 11:17:10 +08:00
Haozhe	188c0fcbf7	refactor(agent-core-v2): decouple kosong from config persistence (#2068) ... * refactor(agent-core-v2): decouple kosong from config persistence - keep the kosong provider/model registries in memory and sync them with config.toml through a new app/kosongConfig two-way bridge: hydrate on startup, push config section changes into the registries, and persist runtime mutations (discovery refresh, OAuth provisioning, default-model pointer changes) back to disk - declare the providers/models/thinking section constants, zod schemas, env bindings, and TOML transforms in a single app/kosongConfig configSection.ts; kosong keeps hand-written types only - pin every schema to its kosong type at compile time via AssertExact<Equal<...>> (_base/utils/typeEquality) - register a transitional auth>kosongConfig exception in the domain-layer checker for the OAuth provisioning flows * chore(agent-core-v2): drop unused IConfigService import from authLegacy * fix(agent-core-v2): reconcile registry with env-pinned default pointers A registry-originated default-model/provider write lands only in the config user layer when an effective overlay pins the section (KIMI_MODEL_NAME pins defaultModel to the reserved env model): the effective value does not move and no change event fires, so the registry diverged from the effective config view — catalog/auth reads reported the user pick while profile resolution kept the pinned model. After persisting, the bridge now re-asserts the effective value into the registry, restoring the pre-refactor behavior where every default-pointer write was arbitrated by the effective view. * fix(agent-core-v2): await persistence before resolving kosong registry mutations ProviderService/ModelService mutations resolved as soon as the in-memory registry updated, while the kosongConfig bridge persisted the change on an unawaited private chain — callers (klient kosong.*, set_default route, OAuth provision, discovery refresh) could observe success before the write reached config.toml, and a restart right after could lose it. - fire registry change events through AsyncEmitter and await delivery in set/delete/replaceAll/setDefaultX, so a mutation resolves only after listeners' waitUntil work completes; loadAll keeps synchronous timing - the bridge hooks its persists into waitUntil, hoists the equality guards into the listeners so config-originated echoes stay fully synchronous, and serializes persists on the chain as before - retry a failed persist with bounded backoff (3 attempts); failures are logged, never rejected to callers, and the in-memory change stands - default-pointer change events now carry an { id } payload (fireAsync requires object events) - add the klient kosong-config stress example covering read-after-write, concurrent bursts, and restart durability	2026-07-23 00:32:10 +08:00
Haozhe	64f053cf46	feat: agent-core-v2 permission/workspace refactors and transcript durability (#2021) ... * refactor(agent-core-v2): extract toolApproval domain from permissionGate - add agent-scoped `toolApproval` domain owning the approval round-trip: builds approval requests, drives the session/approval broker, publishes permission.approval.* events, records session approval rules, and resolves ask continuations - slim permissionPolicy down to the static risk-adjudication chain; drop the dynamic `registerPolicy` mechanism - move harness constraints out of the policy chain into their owning domains as toolExecutor hooks ordered before 'permission': plan-mode guard (plan), swarm batch exclusivity and AgentSwarm approve (swarm), goal-start review (goal), exit-plan review (plan/exitPlanModeReview), and btw deny (session/btw) - delete the now-unused policies (deny-all, plan-mode-guard-deny, plan-mode-tool-approve, goal-start-review-ask, swarm-mode-agent-swarm- approve, agent-swarm-exclusive-deny) - update Permission.md, AGENTS.md, and check-domain-layers for the new domain layout * feat(kimi-inspect): add App Services view for app-scope service reflection - add `services` top-level view (`AppServicesView`) on the NavRail, showing the full-width app-scope Service panel grid; session/agent scopes stay in the Chat view's Inspector - extract shared `ServicePanels` from Inspector and add `methodArgs` helper to build per-method argument editors from channel metadata - support variadic service calls in `panels.ts` (`call(svc, method, ...args)`) - update agent-core-dev skill docs for the toolApproval extraction and the guard/review-off-chain permission design * refactor(agent-core-v2): restructure workspace domains - rename workspaceRegistry to workspace and workspaceLocalConfig to projectLocalConfig - extract id-spelling resolution into the workspaceAliases domain (IWorkspaceAliases) - extract workspace-centric session queries into workspaceSessions (IWorkspaceSessions) - update kap-server routes, klient contracts, and related tests to the new services * feat(transcript): add op-batch sequencing and point-to-point catch-up - wire: transcriptSeqSchema with per-agent batch seq watermark on transcript.reset/ops and the REST transcript response, the transcript_since subscription cursor, and the GET transcript/ops catch-up shape; seq stays optional everywhere so legacy peers fall back to loss-signal-driven refreshes - wire: drop interactionFrame, interactions live on the interaction item in the ops stream - kap-server: TranscriptService assigns consecutive per-agent batch seqs and retains them in a bounded in-memory journal; the transcript_since cursor replays journaled batches instead of a baseline reset, and the baseline reset is now items-empty because history always pages in over REST - kimi-inspect: transcript REST/WS clients track the op-batch watermark and run seq-gap/reconnect catch-up with full-refresh fallback; add the Transcript audit panel (AuditTrail timeline, structural diff, state tree) docked right of the chat - docs: sync AGENTS.md and agent-core-dev skill notes with the new transcript contract * feat(transcript): persist plan revisions and task/interaction facts, add user-messages endpoint - record each ExitPlanMode submission as a versioned plan blob via a reference-only plan.revision op, projected live and cold as a plan.revision marker plus the plan badge (reviewPath, version) - persist task.started/task.terminated (with a bounded output tail) and interaction.request/interaction.resolved ops, and add foldFacts so a cold transcript rebuilds tasks, interactions, todos and goal/plan/swarm meta from the wire journal - add GET /sessions/{session_id}/transcript/user-messages returning all turn-opening prompts grouped per agent * fix(agent-core-v2): anchor external PreToolUse hooks before the permission gate The toolApproval extraction forces the gate to construct early (planService injects it to anchor plan-guard), so the 'permission' hook registered ahead of 'externalHooks' and a policy ask waited on the approval broker before PreToolUse could block, hanging the turn. Fetch the gate first in registerListeners and register the PreToolUse hook with before: 'permission', falling back to appending when the gate is stubbed without its hook. * refactor(agent-core-v2): replace ordered onBeforeExecuteTool hook with veto-event pattern - introduce BeforeToolExecuteEvent with veto/allow/pass/waitUntil statements - add BeforeToolExecuteEmitter with two-pass fire (immediate then deferred) - split readiness work into separate onWillExecuteTool participation event - migrate all domain listeners: permissionGate, plan, goal, swarm, btw, externalHooks, toolDedupe, mcp - remove IAgentPermissionGate force-injection for hook ordering - update docs, tests, and domain-layer check to match * refactor(agent-core-v2): unify veto payload on ExecutableToolResult - veto() and waitUntil factories now carry a plain ExecutableToolResult: isError reads as a denial, anything else as a short-circuit; the block/reason/syntheticResult weak union is gone - add denyToolExecution(reason) helper for the common denial shape, and narrow the fire/authorize return to BeforeExecuteDecision ({ veto } or { executionMetadata }) - narrow the policy 'result' resolution to { kind: 'result'; result } - settle vetoed calls through a single normalize/merge path in the executor * fix(agent-core-v2): pull up IAgentPermissionGate in agent activation The permission gate only subscribes `onBeforeExecuteTool` from its constructor. The veto-event refactor removed the ordering-driven force-injections that used to pull it up, so without an explicit resolution tool execution would run without policy adjudication. * refactor(agent-core-v2): fold systemReminder domain into contextMemory appendTagged - add `appendTagged(content, tag, origin)` to `IAgentContextMemoryService`, storing content pure with a `tag` field on `ContextMessage` - apply the XML tag at projection time in `contextProjector` via the new `tag.ts` helpers (`wrapTag` / `applyTagToContent`) - delete the `systemReminder` domain and migrate all call sites (contextInjector, goal, plugin, prompt, swarm, btw, sessionInit, toolSelectAnnouncements) to `appendTagged` - build toolDedupe reminder strings with `wrapTag` * refactor(klient): merge providers/models/catalog into global.kosong facade Converge three separate facade namespaces (global.providers, global.models, global.catalog) into a single global.kosong facade that exposes two domain concepts: provider (CRUD) and model (read-only view). Add streaming generate() method for direct LLM calls through the facade. - Define ProviderAuth (api-key | oauth), ProviderInput, AnonymousProviderInput, GenerateInput, GenerateParams, GenerateEvent as klient-owned public types - Extend KlientChannel with stream() for AsyncIterable transport - Add streaming IPC protocol (stream/stream_data/stream_end/ stream_error/stream_cancel frame types) - Add StreamingProcedureContract, ScopedStreamCaller, and per-chunk zod validation in the contract layer - Implement generate via dispatcher special-case routing to IModelCatalog.getRequester().request() - Rename events: providers.changed -> kosong.providers.changed, models.changed -> kosong.models.changed, catalog.changed -> kosong.changed - Remove GlobalProvidersFacade, GlobalModelsFacade, GlobalCatalogFacade, and ModelRecord from public exports - Update all tests, examples, and README BREAKING CHANGE: global.providers, global.models, and global.catalog replaced by global.kosong; event names changed; ModelRecord no longer exported. * feat(transcript,kimi-inspect): add tag field to text frames and improve session creation transcript: - add optional `tag` field to TextFrame, textFrameSchema, and HistoryMessage - propagate tag through contextTranscript MutableMessage kimi-inspect: - render tagged frames with violet badge and distinct styling in ChatView - skip cwd prompt for workspace-based session creation in Sidebar - auto-bind default model on new sessions via resolveDefaultModel * refactor(transcript): rename wire/ directory to contract/ The transcript package's REST/WS schemas and event types lived in src/wire/, which collided with the engine's persisted wire.jsonl record vocabulary. Rename it to src/contract/ so "wire" unambiguously refers to wire.jsonl records (foldWireRecordFacts, HistoryWireRecord stay unchanged). - rename src/wire/{schema,events}.ts to src/contract/ - update index exports and test imports accordingly - reword comments: "wire shape" -> "contract shape", "on the wire" -> "in ops" / "in transit" / "on the WS channel" / "transcript API" - events.ts: "transcript frame" -> "transcript event" for WS envelope messages, avoiding confusion with TranscriptFrame - kap-server tests: TranscriptWire/TurnWire/FrameWire/OpsCatchupWire/ UserMessagesWire -> *Contract - update AGENTS.md references * refactor(transcript): rename wire/ directory to contract/ The transcript package's REST/WS schemas and event types lived in src/wire/, which collided with the engine's persisted wire.jsonl record vocabulary. Rename it to src/contract/ so "wire" unambiguously refers to wire.jsonl records (foldWireRecordFacts, HistoryWireRecord stay unchanged). - rename src/wire/{schema,events}.ts to src/contract/ - update index exports and test imports accordingly - reword comments: "wire shape" -> "contract shape", "on the wire" -> "in ops" / "in transit" / "on the WS channel" / "transcript API" - kap-server tests: TranscriptWire/TurnWire/FrameWire/OpsCatchupWire/ UserMessagesWire -> *Contract - update AGENTS.md references * Revert "refactor(agent-core-v2): fold systemReminder domain into contextMemory appendTagged" This reverts commit 55afaa3d96f729c4f73a71f1fbd23d3f6087453b. Restore the systemReminder domain: reminders go back to being baked into message text at write time, and ContextMessage loses the `tag` field (projection-time wrapping is removed with it). * feat(transcript): add wire-equivalent detail, dedupe session events - transcript: add step usage/timing/retry, turn durationMs/error/usage, tool inputText/progress, task resultSummary/error, meta.agent status, a global prompts entity, and the 'hook' marker - kap-server: project the new fields in coreEventMap and suppress transcript-projected session events on connections subscribed to the transcript protocol (live fan-out and cursor replay) - kimi-inspect: mechanical type sync for the new snapshot prompts field * fix(klient): resolve lint errors in ipc channel stream and e2e matrix test	2026-07-22 19:21:56 +08:00
Haozhe	c6291c3ad7	feat(v2-print): align kimi -p run lifecycle with the default engine (#2017) ... * feat(v2-print): align kimi -p run lifecycle with the default engine (13 files) - add applyPrintModeConfigDefaults in agent-core-v2: fill print-mode config defaults (bash task timeout / max steps per turn / subagent timeout all unbounded) into the memory layer, respecting user-set keys - applyPrintBackgroundPolicy: keep the run alive while cron tasks have future fires so their steered turns can run, with an anti-spin guard for wedged ticks; goal/cron waiting shares the steer ceiling budget - default print background mode changed from exit to steer - add task.bashTaskTimeoutS config key and wire it into BashTool's detachTimeoutMs - v2 print runner now forwards turn.step.retrying events to writeRetrying - allow subagent timeoutMs = 0 (no timeout) * test(agent-core-v2): remove duplicate loop config imports in config test	2026-07-22 19:21:39 +08:00
github-actions[bot]	8bf5bacba9	ci: release packages (#1989) ... Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>	2026-07-22 17:24:39 +08:00
Haozhe	8250e590f3	docs(cron): drop references to the non-existent kimi resume command (#2050) ... * docs(cron): drop references to the non-existent `kimi resume` command - point user docs at the real resume command `kimi --session` - reword cron tool descriptions and code comments in agent-core and agent-core-v2 to describe session resume without naming a subcommand * chore: add changeset for cron docs wording fix	2026-07-22 16:59:08 +08:00
Kai	4c763f6763	feat: send prompt-attached videos directly with the prompt (#1999) ... * feat: send prompt-attached videos directly with the prompt Videos attached to a prompt (pasted in the TUI, uploaded in the web UI) previously reached the model only after it opened the file with ReadMediaFile — an extra tool round trip that could leave the video unseen when the model never made that call. They are now uploaded through the model provider's file channel and embedded directly in the user message as the provider-issued reference; ReadMediaFile stays as the fallback and as the model's own way to open video files. - agent-core: new uploadVideo agent RPC and Session.uploadVideo in the SDK; the TUI uploads pasted videos at submit time, falling back to the file-tag form on failure - kap-server: inline file-source video prompt parts at the REST edge and map provider file ids back to local uploads behind GET /files/llm/{llm_id} - kimi-web: play provider-referenced prompt videos after reload/resume * fix: fall back to inline video when the upload channel fails ReadMediaFile only used the provider's video upload channel when one was bound, and surfaced a hard tool error when the upload itself failed — on providers without a files endpoint (or a transient upload failure) the model was told the video could not be read at all. Now a missing or failing upload falls back to delivering the video inline (base64), the same shape providers without an upload channel already get. Both engines (agent-core and agent-core-v2) are fixed. * test: cover prompt video edge cases and failure paths Stress coverage for the prompt video pipeline: - agent-core uploadVideo RPC: extension/magic classification (.txt with video bytes accepted, extension trusted when magic is absent), exact 100MB boundary, directory and nonexistent paths - TUI: mixed per-video outcomes (one inlined, one tag fallback), submission order behind an in-flight upload, queued video messages carrying final uploaded parts - kap-server: per-video provider-id mappings, Range requests through the llm redirect, mapping persistence across a server restart * fix: surface auth rejections from the video upload channel The base64 fallback for a failing video upload must not mask auth rejections: a 401/403 (surfaced as provider.auth_error) drives the credential force-refresh and a clear auth error, while an inline payload would just be rejected again by the next request. Only a missing or broken upload channel (no files endpoint, network/server errors) falls back to inline delivery. * fix: keep the by-design no-hook video error from degrading to inline Main's contract for a provider with no video upload hook is an honest "does not support video upload" tool error — an inline payload would be dropped on that protocol's wire anyway. The base64 fallback now only applies to an upload channel that exists but failed at runtime. The no-hook throw gets a stable type (VideoUploadUnsupportedError) so the two cases are told apart without matching message text. * fix: constrain the llm video route param to a safe alphabet The provider file id is used as a blob-store key, and the node-fs backend joins scope and key into a storage path. An id containing an encoded path separator (%2F) could address a different storage path than the intended llm-video mapping; the route now only accepts the provider-id alphabet. * style(agent-core-v2): move added explanations into top-of-file comment blocks The package's comment convention keeps comments solely in the top-of-file block; relocate the new notes (url-source id pairing, video delivery fallback, VideoUploadUnsupportedError) from functions and schema fields into their module headers. * fix: reject prompt video uploads when the model lacks video input The uploadVideo RPC only checked the provider's upload channel, so an SDK caller on a text-only or unknown-capability model could obtain a valid-looking video_url part the current model is not supposed to accept. The TUI capability guard does not cover this public path, so the agent now gates on video_in itself. * fix: sniff uploaded video bytes before inlining them The inline branch trusted the upload-time content type; now the bytes are sniffed first, and anything magic-confirmed as a non-video kind (e.g. an image mislabeled as video/mp4) falls back to the file-tag form instead of being uploaded. Like the image gate, bytes are authoritative where the container format allows it — an MPEG-PS lookalike still rides the extension, matching ReadMediaFile. * test: keep the generation stub pending until the abort lands An immediately-answered 404 ends the turn (non-retryable) before the test's abort call, racing the cleanup into a 409; the stub now hangs generation until the abort cancels it. * fix: validate prompt file references before mutating session controls A stale file_id failed inside media resolution — after the model, thinking and permission overrides had already been applied — so a rejected prompt still changed the session's controls. File references are now checked up front, keeping failed submits side-effect free. * fix: serialize prompt submissions per session A slow provider video upload let a later text-only request reach the queue ahead of an earlier video one, silently reordering the conversation for REST clients and multiple tabs. Submissions to the same session now chain, matching the ordering the TUI already guarantees locally. * fix: play reloaded provider videos through the authenticated fetch path A video recovered from an ms:// reference carried only the bare redirect URL, which 401s under daemon auth when loaded natively. The attachment now keeps the provider file id (llmFileId) end to end, and AuthMedia fetches the bytes with the Bearer credential through the daemon's llm redirect — the same blob-URL path uploaded files already use. * fix: fence pending video submits against session and model switches A slow paste-upload left the TUI idle, so /new, the session picker or /model could fire mid-upload; the continuation then dispatched the old session's provider reference into the newly selected session or a model that cannot resolve it. The dispatch now re-checks that the session and model are unchanged and asks for a resend instead. * fix: preserve the caller's video path verbatim in uploadVideo Trimming the path changed the filesystem target before validation and upload, so a name that legitimately starts or ends with whitespace resolved to the wrong file; the trim is now only the emptiness check. * fix: forward provider-issued ids on image URL prompt parts too The shared url-source schema accepts id for image and video parts, but only the video path forwarded it, dropping provider-keyed image ids between prompt acceptance and the model request. * fix(web): reconcile inlined video echoes into the optimistic user message The loose user-message matcher counted media parts and <video path> tags but not the [video:ms://…] text shape, so a racing server echo of an inlined upload slipped through as a duplicate user bubble. * fix: queue bash submits behind a pending video upload A bash-mode submit could start while a pasted video was still uploading, recording shell context before the earlier prompt was dispatched and reordering user actions; it now chains behind the upload like normal submits. * test: cast the driver through unknown for the session-switch fence test * fix: validate prompt file references before resolving the prompt agent A stale file_id posted to a fresh or cold session materialized the main agent (registering it in session metadata and igniting agent-scoped services) before the request was rejected. The file check now runs first, so failed submits create nothing and mutate nothing. * fix: key the playback mapping by the id embedded in the reference URL Projections and clients read the provider id from the ms:// URL, not the id field, so an upload that returns an id-less or mismatched reference would inline fine but 404 on playback. The mapping now derives its key from the URL, falling back to the explicit id. * fix: sanitize provider video ids on the write side of the playback map The read route got the safe-alphabet guard, but recordLlmVideoRef still used the provider-returned id verbatim as a blob-store key; a crafted provider response could write the mapping outside the llm-video namespace. Out-of-alphabet ids are now dropped on both put and get. * fix(web): keep recovered provider videos resendable through the edit path The composer reload path only honored fileId and fell back to a bare fetch(url) without the Bearer token, so editing a reloaded provider-video turn dropped the chip after a 401. llmFileId is now threaded through and the bytes are re-uploaded via the authenticated llm redirect. * fix: fall back to inline video for no-hook providers whose wire carries it The by-design no-hook error is only the honest answer when the wire would drop an inline payload anyway (the OpenAI family). Protocols that convert video_url (kimi, anthropic, google-genai, vertex) now take the base64 fallback instead of failing every video read; the registrar computes the flag from the model's protocol. * test: satisfy the full IBlobStore shape in the playback-map stub The tsgo typecheck job rejects a structural stub missing _serviceBrand and list even though plain tsc accepted it. * fix: queue prompt-producing slash commands behind pending uploads Skill activations and plugin commands started their turn immediately while a pasted video was still uploading, so the earlier video prompt queued behind them and user actions ran out of order. Both paths now chain behind inputSubmitChain (and re-check session/model at dispatch) like normal and bash submits. * fix: validate media kinds in the prompt file-reference preflight The preflight only proved a referenced file exists; a real upload used with the wrong kind (e.g. a PDF submitted as video) still passed it and mutated session controls before assertMediaFile rejected the request. The kind assertion now runs up front with the existence check. * fix(web): play sent and recovered videos in the file preview The media preview returned early for every kind except image, so a user-turn video chip's play action was a no-op even with llmFileId threaded through. The preview now handles video: bytes come from the authenticated file/llm fetch into a blob URL and render in a native player. * fix(web): preview recovered videos with the authenticated blob URL The llm re-upload branch fetched bytes with auth but kept the protected redirect URL as the chip preview, which 401s as a native video src; the fetched blob now becomes the preview URL, mirroring the fileId branch. * style(agent-core-v2): move the inline-fallback note into the module header Same package comment convention as before: rationale lives in the top-of-file block, not beside the registration call. * fix: fence delayed bash submits to the originating session The chained bash callback ran runShellCommandFromInput against whatever session was active at dispatch time, so a command submitted in session A could execute in session B's workspace and be recorded there after a mid-upload switch. The originating session is now captured at submit and re-verified at dispatch, like the prompt and skill paths. * fix: emit prompt video telemetry from the agent scope video_upload is an agent-level event requiring ambient agent identity, but the uploader was built with the Core-scoped session view, leaving prompt-upload events unattributable. The route now resolves telemetry from the target agent for the uploader while image compression keeps the session-scoped view. * fix: preserve provider image ids through legacy projections and web mappers The url-source id accepted by the prompt schema was dropped again by the legacy message projection and the web wire mapper, so provider- keyed image references lost their id across messages, snapshots, and undo responses. It now flows through both directions. * fix(web): revoke recovered video blob URLs before dropping attachments A failed llm re-upload removed the attachment without revoking the freshly created preview blob URL, pinning the whole video in the browser blob store until page unload on every failed edit attempt. * fix: serialize foreground slash commands behind pending uploads /compact and /init started their turn while a pasted video was still uploading, so the earlier message landed after them — and compaction summarized the context without it. Prompt-producing builtin commands now share the same queueBehindPendingUploads chain (with the session/model dispatch fence) as skill, plugin, and bash submits. * fix: defer session controls until media preparation succeeds Media resolution now runs before any profile/model/thinking/permission/ denylist mutation, with the uploader resolved transiently from the requested (or currently bound) model — a failed submission leaves the session's controls untouched. A concurrent model switch during preparation is rejected with session.busy instead of enqueueing a reference uploaded for the previous model. * chore: bump the new SDK video upload API as a minor release Session.uploadVideo is new public API surface, not a patch-level tweak. * fix: re-check busy state before draining upload-queued commands A slash command deferred behind a video upload ran the moment the video prompt dispatched, landing on an already-running turn: beginSessionRequest wiped the active turn's live pane, and /init or /compact started on top of it. The deferred callbacks for skills, plugin commands, /compact and /init now re-run the resolver's busy check at dispatch and show the same blocked message the user would get when typing while streaming. * fix: resolve profile-bound models before choosing the uploader A first prompt carrying "profile" without "model" resolved the upload model from the still-unbound alias, so no uploader was installed and every attached video fell back to a tool-read tag even when the configured default model supports provider upload. The transient resolution now mirrors AgentProfileService.bind: an explicit body model wins, a profile bind falls back to the configured default model, and only otherwise does the currently bound alias apply. * refactor: resolve prompt videos at request time inside the engine Move prompt-video delivery out of the submission edge: the TUI submits synchronously with a local file:// part the v1 turn resolves before the message enters history, and kap-server carries an internal kimi-file:// reference the v2 requester resolves against the effective model with an app-scoped upload cache. History keeps the durable local file id, the /messages projection emits structured video parts, and the web plays videos back through the authenticated /files channel - deleting the submit fences, per-session serialization, provider-id reverse mapping, redirect endpoint, and the unreleased SDK upload API. * fix: propagate abort through video upload delivery instead of degrading A turn cancelled mid-upload used to be treated as an ordinary upload failure: v1 fell back to an inline base64 part and appended the degraded message to history, and the v2 resolver memoized the tag fallback for the rest of the agent's lifetime. Both catch sites now check the delivery signal itself - abort rejections vary in shape by provider - and re-throw so cancellation ends the turn (v1, classified as cancelled via the abort reason) or the request (v2, not memoized, so the next turn uploads). * fix: keep the tag form for no-upload providers whose wire drops inline video An OpenAI-family model configured with video_in but no provider upload channel used to receive prompt videos as an inline base64 part - which chat completions rejects and the Responses adapter degrades to an omitted-video placeholder, persisting ~4/3x the file size in history for bytes the model never sees. The prompt path now mirrors the v2 resolver's protocol gate and degrades to the <video path> tag instead; ReadMediaFile's own delivery is unchanged. Also merges the prompt-video changesets into a single user-facing entry. * fix: escape the NUL separator in the video upload cache key The cache-key template literal contained a literal NUL byte instead of the \0 escape, which made Git classify the whole source file as binary - no inline diffs, unreliable text tooling. The escape produces the byte-identical runtime string, so hashed cache keys are unchanged. * fix: check the abort signal before the inline video fallback The no-uploader inline path (and the post-upload-failure fall-through) never consulted the delivery signal, so cancelling a turn while the video bytes were being read still base64-encoded the file and appended the degraded message to history. The inline branch now re-throws the abort reason first, matching the upload catch. * fix: retry transient prompt video upload failures on later steps A generic upload failure used to memoize its tag fallback for the rest of the agent's lifetime, freezing a transient files-endpoint error into a permanently degraded video. The resolver now marks failure-born fallbacks as non-memoizable: the current request keeps the lightweight tag form and the next step retries the upload. Structural outcomes (successful uploads, capability and sniff fallbacks, no-hook inline) stay memoized for step-retry stability.	2026-07-22 13:46:00 +08:00
Kai	ba921ca531	fix: gate always-thinking inference to OpenAI wires, plus catalog review follow-ups (#2036) ... * fix: gate always-thinking inference to OpenAI wires, plus review follow-ups - catalog: strip the inferred alwaysThinking marker on non-OpenAI wires so Claude/Gemini keep their native off; verified against live models.dev - v1 provider-manager: honor per-alias baseUrl on kimi/google-genai/vertexai - thinking: rewrite the PHASE-6 contract comment, drop three dead cannot-disable warning branches, normalize requested effort in v1 to match v2 - tests: input-cap compaction preference in both engines, kap-server WS status cap, v2 legacy status input cap - docs + changeset: new model fields, catalog-refresh behavior, kap-server * fix: strip always-thinking only where the wire encodes a true off The previous gate kept the marker only on the OpenAI wires, which wrongly stripped it from Gemini 3 on the Google wires: its floor is thinkingLevel MINIMAL with suppressed thoughts — still reasoning — so an Off option there would be a lie. The criterion is now the wire's encoding, not its family: strip only on anthropic and kimi, the two wires with a protocol-level `thinking: {type: 'disabled'}` that the catalog's effort list can never show. Verified against live models.dev data: google now marks exactly the gemini-3 family (10), anthropic and moonshotai stay 0. * docs(agent-core-v2): move the strict-validation contract into the thinking.ts file header The scoped guide keeps comments in the top-of-file block only; the function JSDoc shrinks to a short what-it-answers note matching its neighbors. No behavior change.	2026-07-22 13:43:10 +08:00
Haozhe	430cd382a8	refactor(agent-core-v2): drop pass-through methods from AgentRPCService (#2042) ... - trim AgentRPCService/AgentAPI to the 10 methods with real logic (prompt, steer, cancel, undoHistory, setPermission, cancelCompaction, activateSkill, activatePluginCommand, getContext, getTools) - route the klient agent facade to the domain services directly (shellCommand, profile, usage, plan, task) with new wire contracts - keep the test harness ctx.rpc surface unchanged via passthrough adapters - rewire kap-server rpc tests and the kimi-inspect RPC panel	2026-07-22 13:42:01 +08:00
Kai	ec88d352e8	fix: five correctness follow-ups to the catalog metadata work (#2030) ... * fix: five correctness follow-ups to the catalog metadata work - Normalize a configured effort value (case/whitespace) before thinking resolution on both engines, so "OFF" is read as off instead of being sent upstream as an invalid effort. - Clamp a declared input cap to the effective context window in the effective model resolution, so an override lowering max_context_size cannot leave a stale larger cap behind. - Report context usage against the same effective cap (max_input_tokens ?? max_context_tokens) in the SDK getStatus and the v2 legacy status projections, matching the session-event and service surfaces. - Preserve concretely declared per-model endpoints when the override npm is unrecognized, via the same OpenAI-compatible fallback used for top-level entries. - Attribute resolved.maxInputSize and resolved.capabilities .max_input_tokens in the model inspector to their config, override, or clamp provenance. * fix: honor observed context caps and publish the effective cap in v2 status - The overflow-learned provider window is now written into both max_context_tokens and max_input_tokens of the effective compaction context, so the strategy cannot bypass it by re-selecting the raw catalog input cap during overflow recovery. - agent.status.updated publishes max_input_tokens ?? max_context_tokens, matching the other v1/v2 status surfaces. * fix: never mutate config on clamp, cap usage ratio, and refuse proprietary override SDKs - effectiveModelAlias / effectiveModelConfig now build a copy before clamping maxInputSize to the effective window instead of rewriting the caller's config record in place. - context_usage is clamped to 1 in the v2 legacy status, the v1 session service, and the SDK getStatus; the default-model status fallback also resolves the input cap. - Overrides naming known proprietary SDKs (Bedrock, Cohere) are refused before the OpenAI-compatible fallback, matching top-level import behavior. - The inspector attributes a clamped maxInputSize to the clamp even when the raw value came from models.*.overrides. * fix: normalize forced efforts, prefer input cap in WS status, keep raw event ratio - Whitespace-only configured efforts now read as absent, and the KIMI_MODEL_THINKING_EFFORT override is lowercased on both engines. - kap-server's WS legacy status publishes max_input_tokens ?? max_context_tokens like the other status surfaces. - SDK getStatus keeps the ratio unclamped on purpose (>100% is the documented overflow signal on that path); schema-bounded REST status stays clamped to 1. - Pin the provider-observed window beating a declared input cap with a v1 full-compaction regression test. * fix: attribute resolved.maxInputSize provenance and drop test-body comments in v2	2026-07-22 03:13:10 +08:00
Kai	b5efba7abc	fix: consume the model metadata declared by the models.dev catalog (#2015) ... * fix: stop advertising Claude thinking efforts for non-Claude models Models served over the Anthropic protocol whose names carry no Claude marker (e.g. a catalog-imported Kimi K3) no longer inherit the latest Opus effort list, so the model selector stops offering levels the model does not accept. The models.dev catalog import now also parses reasoning_options and records the declared effort levels on the model alias, so K3 offers its real levels (low / high / max). * fix: consume deprecated, override, and input-limit metadata from the models.dev catalog - Models declared status=deprecated in the catalog are no longer offered for import. - Per-model provider overrides on gateway providers (an npm package targeting an Anthropic SDK plus a usable endpoint) now land as alias protocol and base_url, so those models are served over the right protocol and endpoint; overrides without a usable URL are skipped. - A declared limit.input now sizes the context budget instead of the larger total context window (e.g. gpt-5: 272k instead of 400k). The model alias schema gains an optional base_url field (not accepted in overrides) that Anthropic wire resolution prefers over the provider-level base URL. * fix: honor thinking-disable semantics and the OpenAI-compatible fallback in catalog imports - reasoning_options 'none' is the model's off encoding: off_effort flows from the catalog through the model alias to the OpenAI wire providers, so turning thinking off sends 'none' instead of omitting the effort field; models with effort levels but no way to disable thinking are imported as always_thinking and no longer offer an Off option. - Bare Claude family aliases (e.g. sonnet-latest) recover the inferred Anthropic effort profile; v2 comment conventions restored. - Providers whose SDK the catalog does not type now fall back to the OpenAI-compatible wire (with a visible "guessed" note) instead of being refused; imports lacking a usable endpoint ask for one (--base-url on the CLI, a prompt in the TUI). Proprietary SDKs (Amazon Bedrock), unrecognized explicit types, and env-placeholder URLs are refused with a clear reason. * fix: align catalog imports with the reference models.dev consumer - A JSON null tier in declared effort values is now read as the 'none' off-encoding (previously such models were wrongly imported as always-thinking with no way to turn reasoning off). - Alpha-status models are filtered out alongside deprecated ones. - Models whose per-model provider override targets a wire that cannot be expressed per-model (e.g. Claude on google-vertex, whose wire here is Gemini-mode Vertex, or gpt entries on an Anthropic provider) are skipped instead of being imported under the silently wrong protocol. - interleaved: true no longer pins reasoning_content: the provider's default three-field scan is wider and the pinned key only narrowed reasoning parsing for gateways answering with another field name. * fix: require endpoints for Anthropic-compatible catalog imports and honor --base-url - catalogProviderNeedsBaseUrl now covers the Anthropic wire: a non-official Anthropic-compatible vendor without a concrete catalog endpoint (e.g. google-vertex-anthropic) must supply --base-url / the TUI prompt instead of silently falling back to the default Anthropic endpoint. - --base-url now takes precedence over the catalog-declared endpoint, and an empty --base-url is rejected instead of persisting a blank endpoint. * fix: enforce always-on thinking on every wire and refuse Cohere at import A model that declares always_thinking (e.g. a catalog-imported gpt-5) no longer resolves to a dishonest off state via thinking.enabled=false or an SDK/ACP off request: resolution clamps to the model's default effort on every wire instead of letting upstream keep reasoning while the UI reports Off. The Anthropic warn-and-send path for unlisted effort levels is unchanged. Cohere's proprietary SDK joins Amazon Bedrock on the import-refusal list instead of being guessed as OpenAI-compatible. * fix: harden catalog import edge cases - An explicit but unrecognized catalog type is now refused before npm/id inference, so a future catalog protocol is never silently miswired through the OpenAI fallback. - User-supplied --base-url values for Anthropic-wire providers get the same trailing-/v1 normalization as catalog endpoints, avoiding /v1/v1/messages requests. - The TUI import prompt rejects env-placeholder base URLs like the CLI does. * fix: await the floating assertion promise in the catalog add CLI test * refactor: unify catalog import resolution into a single decision function Wire-type inference, the OpenAI-compatible fallback, proprietary-SDK refusal, endpoint adaptation, and the base-URL requirement are now produced together by resolveCatalogImport, one pure resolver consumed by both the CLI and the TUI — replacing the cooperating predicates (inferWireType, isGuessedWireType, catalogProviderNeedsBaseUrl) whose permutations kept producing edge cases. No behavior change. * fix: close configured-off clamp hole, keep inferWireType compat, carry same-wire override endpoints - A configured thinking.effort = "off" no longer bypasses the always-on clamp: it is treated as absent and the model default applies, mirrored on both engines. - The previously public inferWireType stays as a deprecated compatibility wrapper over resolveCatalogImport so existing SDK consumers do not break on a patch release. - Catalog model overrides that stay on the provider's wire but declare their own endpoint now persist it on the alias (and the v1 OpenAI wire branches honor alias-level base URLs like the Anthropic branch). * fix: split total window from input cap and close override/endpoint gaps - max_context_tokens once again means the total context window (used by completion budgeting); a model's declared input limit is tracked as max_input_tokens, which compaction, context-splice and usage-ratio checks prefer — fixing the over-clamping introduced when the input cap was stored as the context budget. - A catalog endpoint declared only as an env placeholder now always produces needs-base-url (official SDK included), so credentials are never sent to the public vendor host by default. - api-only per-model overrides are honored as same-wire endpoint changes; overrides targeting another known but inexpressible wire (e.g. google-genai on an OpenAI gateway) are skipped; same-wire models whose declared endpoint is an unusable placeholder are skipped instead of silently rerouted. * style: drop a function-level comment from the v2 thinking resolver * chore: consolidate the PR's changesets into two user-facing entries	2026-07-22 01:23:31 +08:00
liruifengv	8e0dcf3049	feat(kap-server): add GET /api/v1/oauth/usage for managed account plan usage (#2027)	2026-07-22 00:51:28 +08:00
liruifengv	0a1b5fa00f	fix(agent-core-v2): clear dangling defaultModel and thinking on managed logout (#2026)	2026-07-21 23:50:09 +08:00
Kai	37eda4e59a	feat(config): add env overrides for loop control and background task limits (#1993) ... * feat(config): add env overrides for loop control and background task limits Add three operational environment overrides, resolved as env > config.toml > default in both engines (agent-core and agent-core-v2), matching the existing KIMI_IMAGE_MAX_EDGE_PX / KIMI_SUBAGENT_TIMEOUT_MS pattern: - KIMI_LOOP_MAX_STEPS_PER_TURN overrides loop_control.max_steps_per_turn - KIMI_LOOP_MAX_RETRIES_PER_STEP overrides loop_control.max_retries_per_step - KIMI_CODE_BACKGROUND_MAX_RUNNING_TASKS overrides background.max_running_tasks Invalid values are ignored and fall back to the config value. The v2 engine resolves them through config section env bindings (effective-only, never persisted); the v1 engine resolves them at the consumption point. * fix(config): strip env-bound fields before persisting config writes Environment overrides resolved into the effective config could be echoed back through IConfigService.set/replace (e.g. GET then POST /api/v1/config) and persisted into config.toml, outliving the env var. This affected the pre-existing image / subagent / keep-alive bindings as well as the new loop control and max-running-tasks bindings. Extend ConfigStripEnv with a getEnv parameter and add a shared stripEnvBoundFields helper: while a field's env var is set, writes restore the field's raw on-disk value (or drop it) instead of persisting an echoed env value; when unset, normal writes persist. Register it for the loopControl, task/background, image, and subagent sections. Also fix ConfigService.stripEnv looking up rawSnake by the camelCase domain key; on-disk sections are keyed snake_case. * fix(config): honor binding parsers when stripping env fields on persist An invalid env value (e.g. KIMI_LOOP_MAX_STEPS_PER_TURN=abc) is ignored on the read path but still marked the field env-owned on the write path, so a config write for that field was silently dropped. stripEnvBoundFields now derives the guard from the section's envBindings and skips fields whose env value fails the binding's parse, so invalid env values are ignored on both paths — and the duplicated field/env descriptor list is gone. Also drop function-level comments added beside helpers; agent-core-v2 keeps comments solely in the top-of-file block, so the strip semantics now live in the config.ts / configService.ts headers. * fix(config): re-apply env overlays from the env-free base on every read Two follow-ups from review: - ConfigService.get()/getAll() re-applied section env bindings on the already-overlaid effective cache (and get() mutated it in place), so a valid override degraded to invalid or unset kept serving the stale value until the next reload — and an echoed stale value could then be persisted by a config write. Reads now recompute from a cached env-free validated base, so degraded or removed env values fall back to the file immediately. - stripEnvBoundFields restored env-owned fields from the raw snake sub-object, missing values persisted under legacy keys (e.g. max_steps_per_run). Section stripEnv now receives the env-free, fromToml-normalized raw base, so legacy aliases are honored. * test(kap-server): retry temp-home cleanup in auth tests auth.test.ts removed its temp home with a plain recursive rm, which races late async writers in the server shutdown path and flakes with ENOTEMPTY (seen on main CI and locally on main). Harden the cleanup with the same maxRetries/retryDelay options sessions.test.ts already uses. * chore(changeset): consolidate the env-override changesets into two entries One feature entry (loop/background env overrides, both engines plus the CLI) and one fix entry (env values persisting on writes and sticking after degrade/unset, agent-core-v2 plus the CLI). * fix(config): clear fully-stripped sections and refresh overlay domains on get() Two follow-ups from review: - stripEnvBoundFields returned an empty object when every written field was env-owned, so a section with a non-empty default (e.g. subagent) stored raw = {} and the default stopped applying until the next reload. A fully stripped result now clears the raw section instead. - get(domain) only recomputed env overlays for sections with env bindings, so domains written solely by a ConfigEffectiveOverlay (models / defaultModel from KIMI_MODEL_NAME) kept serving stale cached values after the env changed. get() now derives every non-memory domain from the fresh env-free base, matching getAll(). * test(config): drive the get() overlay freshness test with an inline overlay The previous version imported the model env overlay to activate it, which the CI runners failed to resolve from this test file (both tsgo and vitest, while the same specifier resolves elsewhere — not reproducible locally). An inline ConfigEffectiveOverlay double exercises the same ConfigService contract with a tighter seam and no module dependency. * fix(config): preserve unknown fields on full strip and clone nested env targets Two follow-ups from review: - stripEnvBoundFields cleared the raw section whenever the stripped result was empty, so an env echo write could drop unknown forward-compatible fields from the TOML table. An emptied section now keeps its raw table while the env-free base still holds other fields, and is cleared only when nothing remains (defaults keep applying). - applyEnvBindings reused nested child objects in place, so a nested env binding on a persistable key would mutate the env-free validated base and serve stale values after the env var is removed. Children are now cloned before descending. * fix(config): keep the env-free base when a strip leaves nothing to persist Returning {} for a fully stripped write stored the empty object into the raw layer while the TOML table survived via rawSnake, so the bases diverged; a second echo write then saw an empty base, cleared the section, and deleted the table — dropping unknown forward-compatible fields. When nothing persistable remains, the write is now a no-op for the section (the env-free base is kept as-is), and the section is cleared only when the base is empty. * fix(config): revalidate stripped results so restored raw values never persist A strip may restore on-disk values from the unvalidated raw base (e.g. an env-masked invalid field), smuggling them past the merge-time validation into the stored config: the section would then fail the next buildValidated pass, dropping accompanying valid edits from the runtime while the invalid value stayed persisted. set() now revalidates the stripped result — discarding the parse output so unknown fields survive — and rejects the write instead, matching replace().	2026-07-21 19:49:00 +08:00
7Sageer	a3699dd6aa	feat(kap-server): expose per-tool active state in GET /api/v1/tools (#2005) ... * feat(kap-server): expose per-tool active state in GET /api/v1/tools * Update tools-list-active-flag.md Signed-off-by: 7Sageer <7sageer@djwcb.cn> --------- Signed-off-by: 7Sageer <7sageer@djwcb.cn>	2026-07-21 19:23:15 +08:00
Haozhe	d67a2003ab	feat(kap-server): add GET /api/v1/fs:content endpoint for host files (#2012) ... * feat(kap-server): add GET /api/v1/fs:content endpoint for host files - serve any host file by absolute path as raw content with Content-Type, ETag caching, and single-range support - extract shared file metadata helpers (binary detection, line counting, etag, mime/language guessing) into agent-core-v2 _base/utils/fileMeta - extract pickHeader/parseRangeHeader into kap-server lib/httpRange for reuse across download/content routes * fix(kap-server): resolve lint and typecheck errors in fs:content route Return the reply.send(stream) result through the handler promise instead of using return-await, matching the session fs download route pattern. * fix(kap-server): reject non-regular files in fs:content Device nodes, FIFOs, sockets, and zero-size virtual files (/proc, /sys) would otherwise hang the request or stream unbounded data. Addresses review feedback; also folds fileMeta symbol comments into the header block per the agent-core-v2 header-only comment convention.	2026-07-21 18:53:27 +08:00
7Sageer	92576e4d85	fix(agent-core-v2): reconnect dropped MCP servers on tool call failure (#1991) ... * fix(agent-core-v2): reconnect dropped MCP servers on tool call failure When an MCP transport dies mid-session (e.g. the stdio process exits), the wrapped tool now reconnects the server once and retries the call on the fresh client, so a dropped connection surfaces as a slow call instead of a failed turn. Concurrent failing calls on the same server share a single reconnect, and abort errors bypass the retry. * fix(agent-core-v2): constrain MCP reconnect retries * fix(agent-core-v2): recover MCP tool calls on any transport failure Widen the tool-call recovery trigger from the ConnectionClosed error code to any transport-level failure: an McpError means the server answered (or the request timed out) and is not retried, while raw transport errors (fetch failures, "Not connected" on an already-closed client) now trigger one reconnect-and-retry. This covers remote HTTP/SSE servers restarting and batched calls dispatched after the close was processed. Reconnect through the session-scoped connection manager with join semantics (reconnectAndJoin) instead of a per-agent dedupe map: parallel failing calls across agents now share one reconnect instead of preempting each other through attemptId, and a call that fails after the server was already healed retries on the resolved client without restarting it again. * fix(agent-core-v2): probe MCP liveness before reconnecting on tool call failure Classify an ambiguous tool-call failure by probing the client with a short-timeout ping instead of guessing from the error type: - The server answered (a JSON-RPC error, or a result that failed client-side CallToolResultSchema validation) is rethrown as-is; reconnecting would not change the answer and previously restarted the stdio server on every such call. - The probe succeeding means the failure was transient, so the call is retried once in place instead of paying a full reconnect. - Only a measured-dead transport (the SDK's onclose, or a failed probe) goes through reconnectAndJoin and retries on the fresh client, still capped at one reconnect per call. MCPClient gains ping() for the probe; retries remain at-least-once, now documented in the adapter's module header. * fix(agent-core-v2): keep MCP tools registered on connection failure A server whose transport dies between turns no longer loses its tools: failed/pending status events no longer unregister them, so the next call reaches the adapter and its reconnect-and-retry path heals the connection instead of failing with "tool not found". * Improve MCP server connection handling on tool calls A server whose connection dies between turns no longer loses its tools from the tool list — they stay available and the next call heals the connection instead of failing with 'tool not found'. Signed-off-by: 7Sageer <7sageer@djwcb.cn> --------- Signed-off-by: 7Sageer <7sageer@djwcb.cn>	2026-07-21 17:56:06 +08:00
liruifengv	e45832398d	fix(tui): keep long sessions responsive and bound resumed history (#1976) ... * fix(pi-tui): reuse processed lines across frames in the renderer Each frame previously re-truncated, re-normalized, and re-compared every transcript line, so steady-state frames (spinner ticks, streaming flushes) cost O(total lines x chars) and pegged one core in long sessions. Keep the previous frame's raw lines, processed output, and per-line kitty image ids; a line whose raw string reference is unchanged reuses its processed output verbatim, and image-id consumers read the cache instead of re-scanning text. * fix(tui): stop tree-wide transcript invalidation on structural updates Grouping a second Read/Agent call, removing swarm progress, finalizing an MCP status row, replay tool-call removal, and /undo each invalidated the entire transcript tree, forcing every mounted message to re-render (markdown lexing + code highlighting) on the next frame. The container's render cache already validates per child by reference, so structural child-list changes are picked up without a tree-wide invalidate; reserve it for global style changes such as theme switches. * fix(tui): fold older assistant messages into the turn step summary Step merging only collapsed thinking/tool steps, so assistant text blocks accumulated without bound inside a turn (hundreds of markdown components in a single long turn, all re-processed every frame). A running turn now keeps its last 20 assistant messages (KIMI_CODE_TUI_KEEP_RECENT_ASSISTANT), and a finished turn folds down to its conclusion tail of 2 (KIMI_CODE_TUI_KEEP_RECENT_ASSISTANT_COMPLETED); older ones collapse into the step summary line with a message count. Entries are kept, so expand behavior is unchanged. * fix(session): bound resumed history to the most recent user turns Resume used to return every agent's full replay over the RPC boundary (a long session can reach ~100MB, serialized and parsed several times on an in-process call), while the TUI only renders the last 10 user turns. The resume payload now accepts an optional replayTurnLimit, the turn-boundary predicate moves into agent-core as the single source of truth (limitAgentReplayByTurns, re-exported through the SDK), and the CLI passes its existing 10-turn limit so resume transfers just the tail. * fix(session): count goal continuation rounds as replay turns Replay turn boundaries only matched real user input, so a 100-round goal (a handful of user prompts plus 100 system-trigger continuations) fell entirely inside the 10-turn replay window and resumed by rendering the whole run from the start. The goal driver already fires one synthetic prompt per goal turn and counts those as turns itself, so replay trimming now treats goal_continuation prompts as turn boundaries and keeps the most recent 10 rounds. The continuation prompt is model-facing and hidden live; replay no longer renders it as a user bubble either, while still advancing the replay turn so each round groups separately. * fix(tui): keep replay turn folding when goal continuations are hidden Suppressing goal continuation bubbles removed every turn-boundary component from goal-session replays, so mergeAllTurnSteps found no turn edges and skipped step/assistant folding entirely — a single oversized goal round still mounted all of its tool cards and assistant messages. Mount an invisible ReplayTurnBoundaryComponent for each hidden continuation: it renders zero lines but keeps the turn edges discoverable to folding and window trimming, matching replay trimming's per-round turn semantics. * chore: consolidate and simplify changesets	2026-07-21 15:20:12 +08:00
Haozhe	74da87a457	feat(kap-server): broadcast agent.created/agent.disposed session events (#1997) ... * feat(kap-server): broadcast agent.created/agent.disposed session events - emit durable agent.created / agent.disposed facts from the SessionEventBroadcaster lifecycle callbacks, ahead of the agent's own events, and let them bypass per-subscription agent allowlists - add the agent.created / agent.disposed wire types and zod schemas - stamp disposedAt on the transcript roster entry via TranscriptStore.markDisposed so REST consumers can tell a dead agent from a live one * chore: add changeset for agent lifecycle events	2026-07-21 15:18:59 +08:00
7Sageer	ce0e3ceb04	feat: support custom agent files (#1735) ... * feat: support custom agent files Discover Markdown+frontmatter agent definitions from user, project, and configured directories; merge them into a session-level profile catalog (priority: builtin < user < extra < project < explicit). Custom agents work as subagents via the Agent tool and as the main agent. - agent-core-v2: new agentFileCatalog domain (discovery/parsing/profile factory, mirroring skillRoots/parseFrontmatter) and sessionAgentProfileCatalog merged view; AgentProfile.tools becomes optional (undefined = all tools) and gains disallowedTools deny list, evaluated in profileService.isToolActive and persisted in session wire records so resume keeps the gate even if the file is gone - CLI: restore --agent/--agent-file for the v2 print runner (KIMI_CODE_EXPERIMENTAL_FLAG); the v1 TUI rejects them with a clear v2-only error - kap-server/protocol: optional profile field on prompt submission with first-bind semantics (same-name no-op, different name rejected) - docs: custom agents section (en/zh) + config/CLI references + changeset * chore: shorten custom agents changeset * fix(agent-core-v2): stringify caught errors in agent catalog log calls * docs: drop v2 engine notes from custom agents docs * fix: align custom agent binding semantics across engine and edges - agent-core-v2: bind now owns the first-bind invariant — switching profiles after bind throws profile.already_bound (checked again in the synchronous segment before the first wire dispatch, so concurrent binds cannot both pass); unknown names throw profile.unknown; same-name rebinds keep the persisted thinking effort. - kap-server / CLI: edges degrade to error mapping / same-name no-op instead of their own divergent guards. - agent files: reject non-string mode values, honor disallowedTools in the append-mode Skill probe, pass --agent-file through unresolved so the engine can expand ~, reject empty --agent-file values. - session catalog: ready is recoverable via reload() after a fatal source failure, and agent-file discovery is kicked at session materialize so resumed sessions see file agents from the first turn. - docs: first-bind semantics, name: agent override, tools: [] meaning, --agent-file last-wins. * fix: tighten custom agent behavior * fix: address custom agent review findings * fix(agent-core,agent-core-v2): keep active-tool wire records replayable by v1 Binding a profile without a tool allowlist (the default for file-defined custom agents) persisted a tools.set_active_tools record with no `names` key. v1 clients discover v2 sessions through the shared session index and replay newer wire versions without migration, so the record crashed v1 resume with a TypeError and wedged the session permanently. The v2 engine no longer writes the record when the base set is already "every tool active" (its absence encodes the same state), and v1 replay now skips records that lack `names` as defense in depth for wires already written by preview builds. A same-name rebind that resets an allowlist to "all tools" has no v1-safe encoding and is left as a documented gap (no production caller today); a future tools.reset_active_tools Op is safe because v1 silently no-ops unknown record types. * fix(agent-core-v2): tolerate unreadable directories in agent-file discovery A single unreadable subdirectory (EACCES anywhere in the scanned tree) previously aborted the whole discovery pass, zeroing every agent of that source on every session start with one path-less warning. The walker now skips-and-warns per directory below the root (mirroring the skill discovery it parallels), root-level failures are isolated per root so one bad root no longer takes its siblings down, and only a genuinely transient whole-fs outage (os.fs.unavailable) still propagates so the session catalog keeps its previous contribution. Source-level warnings now name the offending path, and repeated skip warnings are capped with a summary that samples the suppressed paths. Also consolidates the path primitives (~ expansion, base-relative resolution, realpath type probes) shared by the root resolvers, the walker, and the explicit-file source into agentFileCatalog/paths.ts, and tightens parser diagnostics: frontmatter null is treated as absent, and a present-but-wrong-typed name/description reports a type error instead of "missing". * fix(agent-core-v2): warn when a same-name builtin suppresses a file profile A directory-discovered agent file colliding with a builtin profile without override: true was silently dropped at merge time. The suppression now logs a warning naming the profile and the opt-in. * refactor(agent-core-v2): pass skillActive explicitly to renderSystemPrompt The third parameter was a full tool list used only for includes('Skill'), which forced the agent-file profile factory to answer a boolean question with sentinel lists. The template now takes an explicit skillActive flag; a skillActiveFor helper keeps builtin call sites derived from their tool arrays. * refactor(agent-core-v2): fall back to the configured default model in bind BindAgentInput.model is now optional: the engine resolves a missing model against the configured defaultModel and throws model.not_configured when neither is set, so edges no longer each re-implement the fallback. * fix(agent-core-v2,kap-server): reject unsupported thinking atomically at first bind A REST prompt carrying profile + an unsupported thinking effort bound the session first and failed setThinking after, wedging the session on an identity the user never successfully used. The effort is now validated up front when the caller marks it as an explicit request (strictThinking): the bind rejects before any await or state mutation, and the requested effort rides along in the bind instead of a separate setThinking. Internal spawn/fork paths pass inherited thinking without the flag and keep the previous clamp behavior — a persisted effort that drifted out of the model's support list must not break subagent spawning. The route's now-redundant model fallback is dropped in favor of the engine-side default. * fix(agent-core-v2): await the agent profile catalog at session materialize The catalog's ready promise was only kicked, so a resumed session's first turn could render the Agent tool description without the file-defined agent types. Discovery is local-fs and cheap, so materialize now awaits it; ready only rejects for a fatal explicit-source error, which is exactly the case that should fail fast. A failure there now also removes and disposes the half-materialized handle instead of leaving it registered in the session cache. * test: cover the --agent-file fatal path and tidy profile registration hygiene The v2 print CLI now has a test asserting an invalid --agent-file fails before any turn. The denylist profiles in binding.test.ts register in a beforeAll (idempotent, scoped to the describe's run window) instead of at module scope during collection. * docs: align custom agent docs with v2-engine gating --agent/--agent-file are rejected without the v2 engine, so restore the requirement in the Agents and Command Reference pages (and use KIMI_CODE_EXPERIMENTAL_FLAG=1 in the examples), note that tool lists only shape model-visible disclosure (permission rules are the enforcement layer), and remind authors of delegation-bound agents to state the handoff contract in the prompt body. * test(agent-core-v2): revert unrelated style churn in fs/workspace tests Keep these files' diff limited to the realpath fakes the feature needs; the lint-preference rewrites belong to a separate cleanup. * feat(agent-core-v2): add permanent system prompt override via SYSTEM.md Read $KIMI_CODE_HOME/SYSTEM.md on every startup and inject it as the default main-agent profile (name "agent", override: true), replacing the builtin default system prompt while inheriting builtin tools and description. Missing or empty files are ignored; unreadable files warn and fall back to the builtin profile. The body supports variable substitution (${skills}, ${agents_md}, ${cwd}, ${cwd_listing}, ${os}, ${shell}, ${now}); unknown variables pass through verbatim. Priority: --agent-file / --agent / project override > SYSTEM.md > same-name user-scope scan files. * feat(agent-core-v2): gate tools globally and accept session disabledTools Add a [tools] config section: "enabled" acts as a global allowlist (empty = unconstrained), "disabled" as a denylist applied on top, both intersected with the active profile's policy in isToolActive (mcp glob supported). Plumb a session-persistent disabledTools parameter through the stack: v2 RPC PromptPayload, REST "disabled_tools" (protocol and kap-server parallel schemas), klient contract/facade, and node-sdk. The server applies it via profileService.setSessionDisabledTools, which replaces the client-owned denylist, keeps the profile's own deny, persists across resume, and rejects calls before a profile is bound with profile.not_bound (mapped to 40001). v1 core-api gains a type-only field and ignores it. * fix: enforce session tool policy across agents * fix(agent-core-v2): enforce tool policy at execution * fix(agent-core-v2): align subagent tool descriptions with policy * fix(agent-core-v2): harden custom agent policy state * fix(agent-core-v2): harden custom agent lifecycle * refactor(agent-core-v2): persist profile binding in a single profile.bind record * fix(agent-core-v2): skip unreadable paths during agent file discovery * fix(agent-core-v2): exempt select_tools from the executor policy guard - share one composed profile/global/session tool-policy evaluation between the executor gate and prompt rendering instead of two verbatim copies - tolerate context-build failure in system prompt refresh instead of rejecting callers (config watcher void-fire, session policy fan-out) * test(agent-core-v2): resolve profile and tool-policy SUTs by interface - drop the Object.assign patching of tool-policy methods onto the shared profile service; rename describes so the SUT ownership is accurate - classify profile.bind as v2-only with the accepted v1-replay tradeoff documented, un-red the wire vocabulary guard test - cover the select_tools guard exemption with an executor-level test * fix(agent-core-v2): enforce explicit select_tools policy * feat(agent-core-v2): accept Claude-style tool lists and rename agent-file mode to promptMode * feat(agent-core-v2): unify prompt templating on ${var} - Replace the nunjucks renderer with a single ${var} regex renderer (unknown placeholders pass through verbatim) and drop nunjucks from agent-core-v2. - Merge the variable tables into one catalog shared by the builtin system.md, SYSTEM.md, and agent file bodies; adds additional_dirs_info plus code-composed blocks (windows_notes, additional_dirs_section, skills_section). - Replace the agent-file promptMode field with ${base_prompt}: bodies are always rendered as templates, and ${base_prompt} expands to the effective default profile prompt (honoring the SYSTEM.md override). - Migrate the builtin system.md, goal reminders, compaction instruction, and tool description templates to the same syntax. * docs: complete agent priority chain and link SYSTEM.md precedence * test: fix invalid custom agent fixture * fix(cli): reject multiple agent selectors * feat(agent-core-v2): add subagents allowlist to agent files * fix(agent-core-v2): persist the subagent allowlist in the profile binding The delegation allowlist now rides the profile.bind record like the tool denylist, so a resumed session keeps enforcing it even when the source agent file was deleted or changed. Agent/AgentSwarm resolve the caller's allowlist from the persisted binding data instead of looking the profile up in the live catalog. * feat(agent-core-v2): warn on tool patterns that never match Profile bind/apply and [tools] config changes now statically flag entries that can never activate anything — wildcards without the mcp__ prefix (a bare * in an allowlist disables everything, in a denylist nothing), incomplete mcp__ literals, and names no registered or builtin-profile tool has — via a tool-pattern-no-match warning event, once per pattern, instead of letting the tool set silently shrink. The known-name vocabulary is the live registry plus literal names from the builtin profiles, so flag-gated tools stay known and a typo in one agent file cannot legitimize the same typo in another. * docs: align --agent-file docs with the single-selector CLI The flag accepts exactly one file and conflicts with --agent, but the docs still described the earlier repeatable, composable design. * docs: note the agent-file trust model and never-matching tool patterns Spell out that project-scoped agent files can replace the default main agent's whole system prompt (unlike AGENTS.md reference injection), and list the three tool-pattern shapes that never match and now raise a warning. * chore: slim changeset wording to user-facing language Drop wire record names, enforcement mechanics, and template syntax from the entries; split the v1 resume fix into its own patch changeset; add a patch entry for the tool-pattern warnings. * chore: shorten changeset entries to one-line summaries * Delete .changeset/v1-resume-v2-sessions.md Signed-off-by: 7Sageer <12210216@mail.sustech.edu.cn> * test: drop class-instance spread in sessionLifecycle test stub * chore: clear the comments --------- Signed-off-by: 7Sageer <12210216@mail.sustech.edu.cn>	2026-07-21 15:01:43 +08:00
7Sageer	e070a580f0	feat(telemetry): emit turn_id and agent_id on turn and tool events (#1675) ... * fix: emit turn_id on turn_started/ended/interrupted telemetry The turn lifecycle telemetry events (turn_started, turn_ended, turn_interrupted) never carried the turn id, while tool_call and tool_call_dedup_detected did. Any analysis correlating a turn's start, end, or interruption back to its tool calls had nothing to join on. Add turn_id (already in scope) to all three track() calls, matching the existing key/value convention used by tool_call_dedup_detected. Update the strict turn_started/turn_interrupted assertion to cover it. * fix(agent-core-v2): emit turn_id on turn_started/ended/interrupted telemetry Port the v1 fix to agent-core-v2: turn lifecycle telemetry events (turn_started, turn_ended, turn_interrupted) carried no turn id while tool_call did, leaving nothing to correlate a turn's start, end, or interruption back to its tool calls. Add turn_id to the three event interfaces, the telemetry registry property docs, and the three track2() calls in AgentLoopService, matching the existing ToolCallEvent key convention. Extend the turn telemetry assertions in loop.test.ts to cover it. * fix: emit turn_id on tool_call telemetry * docs(agent-core-v2): clarify turn_id is a per-agent index in the telemetry registry * feat(telemetry): add agent_id to turn and tool events turn_id is a per-agent counter, so it collides across the main agent and subagents within a session. Emit each agent's scope id as agent_id on turn_*, tool_call, tool_call_dedup_detected, api_error and subagent_created (plus parent_agent_id) so events become attributable via (session_id, agent_id, turn_id). * docs(changeset): cover turn_id emission alongside agent_id * feat(telemetry): add agent_id to agent-level settings events * feat(telemetry): link v2 events across agents, turns, and tool calls - subagent_created: parent_tool_call_id, so a child run joins to the tool call that launched it - permission_policy_decision / permission_approval_result: agent_id, turn_id, tool_call_id - plan_submitted / plan_resolved / plan_enter_resolved, context_projection_repaired: agent_id - compaction_finished / compaction_failed: agent_id + optional turn_id - cron_scheduled / cron_deleted: optional agent_id of the scheduling agent - api_error: turn_id + request_kind, so compaction request failures are distinguishable from turn requests - tool_call_repeat: agent_id + optional turn_id; tool_call_dedup_detected stops fabricating turn_id: 0 outside a turn - background_task_created/completed: task_id on both, unified kind vocabulary ('process' replaces the legacy 'bash' alias on created) - agent lifecycle: auto-assigned agent-N ids now skip ids persisted by previous runs, so a resumed session cannot reissue agent-0 and collide with earlier telemetry * fix(agent-core-v2): preserve telemetry turn attribution * chore: split telemetry changeset into two logical changes * refactor(agent-core-v2): bind agent telemetry context at scope * test(agent-core-v2): fix telemetry assertions for scope-bound agent_id - undoHistory/goal: include the injected agent_id in exact property assertions - rpc-events: assert on the shared records array instead of a track2 spy, which the scoped telemetry view bypasses * refactor(agent-core-v2): bind agent identity ambiently in telemetry - TelemetryService.withContext now returns a lightweight forwarding view: transport state (appenders, enabled flag) stays with the App-scoped root, so views created before an appender attaches no longer silently drop events, and enablement changes apply to every live view. - Agent-scope events register with defineAgentTelemetryEvent and compose the centrally declared AgentTelemetryEventContext (agent_id) into their wire schema; business payloads and call sites stay free of agent_id, enforced at compile time and by the registry convention test. - image_compress/image_crop stay plain events: the kap-server prompt routes emit them through a session-scoped view without agent identity. - v1 subagent_created gains parent_tool_call_id for parity with v2. - Add a lifecycle test asserting each agent scope's telemetry view binds its own agent id. * Delete .changeset/subagent-id-reuse.md Signed-off-by: 7Sageer <12210216@mail.sustech.edu.cn> --------- Signed-off-by: 7Sageer <12210216@mail.sustech.edu.cn>	2026-07-21 14:48:09 +08:00
Kai	a8f1ca3f10	feat(acp): support selecting thinking effort levels (#1992) ... * feat(acp): support selecting thinking effort levels The ACP thinking config option was a binary on/off toggle mapped to the model's default effort, so ACP clients (e.g. Zed) could not pick a concrete level even though the engine and the SDK fully support support_efforts granularity. - Advertise one select row per declared effort level (off + levels); boolean models keep the legacy off/on rows, always-thinking models drop the off row. - Track the session thinking state as an effort string and reconcile it with the engine-normalized value read back from session status after model/thinking changes. - Validate incoming levels against the current model's catalog and reject unknown ones with invalid_params before any SDK call; keep the legacy 'on' value as an alias for the model default effort. * test(acp): avoid spreading the Session stub in the reconciliation test	2026-07-21 13:48:50 +08:00
Haozhe	6dd4fd3368	refactor(agent-core-v2): rebuild the model wire layer on the kosong architecture (#1970) ... * refactor(v2): land kosong contract layer (L0 wire contract) * refactor(v2): land kosong protocol layer (L1 traits and base registry) * refactor(v2): land kosong provider layer (bases, trait composers, kimi definition) * refactor(v2): land kosong model and catalog layers * refactor(v2): migrate engine callers to kosong, drop old llmProtocol layer * test(v2): migrate agent-core-v2 tests and harness to kosong * refactor: sync peripheral packages to kosong architecture * refactor(v2): replace provider dialects with per-protocol definitions * refactor(v2): merge the vertexai protocol into google-genai via providerOptions * refactor(v2): remove vendor-name gates outside the kosong layer * feat(v2): add model resolution inspection and connectivity ping * refactor(v2): remove the unused platform config layer * test(klient): pin invalid-input behavior across providers in e2e * refactor(agent-core-v2): merge kimi traits, split bases by protocol - merge the seven kimi trait modules into kimi.contrib.ts as two trait objects: kimiOpenAITrait (all native-transport hooks) and kimiAnthropicTrait (thinking only) - move bases/* implementations into per-protocol directories (openai/, anthropic/, google-genai/) and rename openai.contrib.ts to openai-legacy.contrib.ts - add per-directory index.ts registration barrels (import = registration) and exempt them in check-domain-layers.mjs alongside *.contrib.ts * refactor(agent-core-v2): reorganize model and request-layer types - consolidate shared model types (ModelOverrides, CompletionBudgetConfig/Params, ResolvedModelAuthMaterial, ThinkingDefaults/ModelThinkingMetadata) into kosong/model/model.types.ts and drop modelOverrides.ts - rename L2 request types to the ModelRequest* prefix: LLMEvent -> ModelRequestEvent, LLMRequestInput -> ModelRequestInput, LLMCallParams -> ModelRequestParams - extract ModelRequestTiming to replace three duplicated copies of the stream-timing shape - rename L3 llmRequester types with the Agent prefix to match IAgentLLMRequesterService (AgentLLMRequestOverrides/Finish/Task/Source/ PartHandler/LogFields) - delete the unused LLMRequestParams type * refactor(agent-core-v2): fold kosong/catalog into kosong/model - merge IModelCatalogService's enumeration surface (listModels / listProviders / getProvider / setDefaultModel and the wire shapes) into IModelCatalog; delete kosong/catalog/modelCatalog.ts - move the remote refresh path to the new IProviderDiscoveryService (discovery.ts + discoveryService.ts, renamed from catalog/modelCatalogService.ts) - relocate configSection / errors to discoveryConfigSection.ts / errors.ts; DEFAULT_MODEL_SECTION now lives in kosong/model/model.ts - drop the L3 catalog layer from check-domain-layers.mjs - kap-server routes / refresh scheduler / channelRegistry follow the split; klient renames the contract modelCatalogService to modelResolver and adds providerDiscovery; kimi-inspect reads via IModelCatalog - modelRequesterImpl: drop the streamedAnyPart backfill, onMessagePart already delivers every part - tests: remove the app/modelCatalog and kosong/catalog suites, add the kosong/model catalog and discovery suites * feat(klient): trim trailing undefined args and add boundary smoke probes - add trimTrailingUndefined helper so optional trailing args no longer cross the wire as null in http/ipc transports, which defeated server-side default parameters - add model-requester-boundary smoke probe for ChatProvider error wrapping behavior against real config and a local stub - add kimi-select-tools smoke probe verifying the kimi-only wire encoding of dynamic tool declarations - extend smoke.ts with a models set/get/delete round-trip, catalog list assertions, and update AGENTS.md with the new scripts * fix(agent-core-v2): declare openai chat hooks as function properties Method-shorthand members on OpenAIChatCompletionsHooks tripped typescript-eslint(unbound-method) at every extraction site (`const hook = this._hooks?.convertMessage` and friends), failing the repo-wide lint job. Every implementation is a plain closure composed by openaiHooks.ts, so declare the members as function-typed properties, which matches the actual semantics and clears the four errors. * fix: repair stale references surfaced by the origin/main rebase - agent-core-v2: point vacuousContent's ContentPart import at kosong/contract - kap-server: rewrite the transcript test seed as IModelCatalog (IModelResolver is gone) - klient: inline onceEvent/waitFor after the http transport helpers were dropped - kimi-inspect: remove useLiveEvent from ModelCatalogView; catalog polls on a slow interval * chore: downgrade the kosong architecture changeset to patch	2026-07-21 13:03:54 +08:00