Added a builtin plugin that you can open from the chat input plus button menu, which shows you a list of skills that you can directly activate in the current context/project.
Default configs allow users to start over with skills already active, instead of losing time and tokens asking Agent Zero to do it.
Update prompt for manual skill selector
add thumbnail for _skills builtin plugin
- Add FAISS index integrity checks using a SHA-256 sidecar (`index.faiss.sha256`) and write hash on save.
- Harden `memory_load` filter evaluation with input validation (allowlist + length cap) and `simple_eval(..., functions={})`.
- Add score-preserving similarity search and use real relevance scores in consolidation (including best-score dedupe by memory id).
- Prevent utility-model context overflows by truncating memorize input history for fragments and solutions.
Bug: not tool_request.get("tool_args") evaluates True for empty dict {}
causing ValueError crash on any tool call with no arguments.
Scheduler list_tasks, health checks, etc all broken.
Fix: Changed to existence check ("tool_args" not in tool_request)
Co-authored-by: Agent Zero <agent@zero>
Tool execution no longer waits for the full streamed assistant text. We now detect the first explicitly closed top-level JSON object, freeze that snapshot as the canonical tool request, and stop the model stream there for dispatch.
To make that safe, DirtyJson completion semantics are tightened so completed=true only means the root object was explicitly closed, not that parsing hit end of file. I also restricted the new extraction path to object roots only, since tool calls are always brace-delimited objects, and added tests for parser completion and early stream stop.
Track parsing depth via _pop_stack() helper. Exposes a 'completed' flag that signals when the root JSON structure is fully closed, allowing stream consumers to break early instead of waiting for irrelevant tokens.
Trying to steer the model into treating the final curly brace of a JSON obj as EOS token. Also defaulting to concise responses, expanding only when needed.
Restore main.communication, main.solving, main.tips and tool.response, which made the model dumber. For some reasons the drawback was more visible with frontier LLMs.
restore builtin plugins and agent0 profile
`find_plugin_dir` can return `None` if a plugin cannot be found. Passing
this null value to `files.get_abs_path` caused crashes during config
retrieval. `get_plugin_config` and `get_default_plugin_config` now check
for a valid directory and return early if it is missing.
Add the always-enabled `_discovery` plugin to turn the welcome screen into a discovery surface for the Plugin Hub and A0 integrations.
Includes a hero card plus Telegram, Email, and WhatsApp feature cards, with persistent dismiss/restore state, CTA routing to plugin config screens, and self-contained placeholder artwork. Implemented entirely through the existing WebUI extension mechanism with no core welcome-screen changes.
stores cleanup
layout polish and onboarding integration
Move feature card titles beside thumbnails for better space efficiency
and visibility. Restructure card markup and styles to support a fluid
grid layout and horizontal alignment.
Integrate discovery cards into the final onboarding step via a new
'onboarding-success-end' extension point, ensuring new users see
extension opportunities immediately after setup.
Hide discovery cards on the dashboard while the missing API key
onboarding banner is visible to reduce UI noise and user confusion during initial config.
update discovery card initialization and loading logic
Enhance the discovery store to fetch cards from the API, improving the dynamic loading of discovery cards based on user context. This change optimizes the user experience by ensuring relevant cards are displayed immediately after onboarding and when modals are closed.
And on top of that, there's a proper backend for these new cards.
- Extract UI server setup into UiServerRuntime class with modular initialization
- Move environment configuration, route registration, and transport handlers to helpers/ui_server.py
- Add released_at timestamp tracking for git tags and branch heads across update system
- Implement get_current_major_main_latest_info to find latest same-major version on main branch
- Add major_upgrade_versions and main_branch_latest fields to update info payload
- Remove
Promote the a0_small prompt work into the shipped default stack.
Token count for default prompt, accounting for tools, plugins, secrets placeholders, inactive project, etc: 2996 tokens.
- migrate compact prompt fragments into core prompts, plugin-owned prompts, and the agent0 overlay
- remove the obsolete a0_small profile and rename the tool-call knowledge reference to a generic framework file
- keep a small set of high-value JSON examples while preserving key guardrails for tool names, tool args, and method-style calls
- bake balanced conciseness into default response behavior without changing the initial greeting
- drop the call_sub prompt-side loader and keep call_sub guidance self-contained
- restore essential secrets/security guidance, preserve inactive-project messaging, and rebalance the A2A prompt
- add a regression test that verifies the assembled default agent0 prompt stays under the token budget and retains critical guardrails
Ensure printed output and HTML logs are safe by importing and applying sanitize_string, opening log files with utf-8 and errors='replace', and sanitizing text before writing. Add tests to verify lone surrogate characters are replaced and that logging won't crash on invalid Unicode. In the plugin installer UI, introduce POPULAR_PLUGIN_MIN_STARS and centralize popularity checking in _isPopularPlugin, using it for filtering and counts.
Created a new prompt set for small models.
A0 now starts with 2.5k tokens usage, not 10k.
improve small model tool policy
prompts: knowledge with tool call examples
- Adds knowledge examples for tool-call shapes and memory usage (workaround to not have examples in tool prompts)
- Adds a small prompt hint to consult memory examples when tool args are unclear
stricter small-model guardrails
Canonical JSON example, exact tool-name matching, method suffix enforcement, British spelling for behaviour_adjustment, boolean guidance for call_subordinate.reset, and the “don’t improvise tools after an error” rule are the new additions here.
Trying to make
