* WIP
Refactor resolving mode_names, should be done once at MCP call boundary
Pass around model context instead
Consensus tool allows one to get a consensus from multiple models, optionally assigning one a 'for' or 'against' stance to find nuanced responses.
* Deduplication of model resolution, model_context should be available before reaching deeper parts of the code
Improved abstraction when building conversations
Throw programmer errors early
* Guardrails
Support for `model:option` format at MCP boundary so future tools can use additional options if needed instead of handling this only for consensus
Model name now supports an optional ":option" for future use
* Simplified async flow
* Improved model for request to support natural language
Simplified async flow
* Improved model for request to support natural language
Simplified async flow
* Fix consensus tool async/sync patterns to match codebase standards
CRITICAL FIXES:
- Converted _get_consensus_responses from async to sync (matches other tools)
- Converted store_conversation_turn from async to sync (add_turn is synchronous)
- Removed unnecessary asyncio imports and sleep calls
- Fixed ClosedResourceError in MCP protocol during long consensus operations
PATTERN ALIGNMENT:
- Consensus tool now follows same sync patterns as all other tools
- Only execute() and prepare_prompt() are async (base class requirement)
- All internal operations are synchronous like analyze, chat, debug, etc.
TESTING:
- MCP simulation test now passes: consensus_stance ✅
- Two-model consensus works correctly in ~35 seconds
- Unknown stance handling defaults to neutral with warnings
- All 9 unit tests pass (100% success rate)
The consensus tool async patterns were anomalous in the codebase.
This fix aligns it with the established synchronous patterns used
by all other tools while maintaining full functionality.
🤖 Generated with [Claude Code](https://claude.ai/code)
Co-Authored-By: Claude <noreply@anthropic.com>
* Fixed call order and added new test
* Cleanup dead comments
Docs for the new tool
Improved tests
---------
Co-authored-by: Claude <noreply@anthropic.com>
New tool to list all models `listmodels`
Integration test to for all the different combinations of API keys
Tweaks to codereview prompt for a better quality input from Claude
Fixed missing 'low' severity in codereview
Instead of creating new OpenRouterModelRegistry instances multiple times
per tool (4x per tool during schema generation), we now use a shared
class-level cache in BaseTool. This reduces registry loading from 40+ times
to just once during MCP server initialization.
The optimization:
- Adds _openrouter_registry_cache as a class variable in BaseTool
- Implements _get_openrouter_registry() classmethod for lazy loading
- Ensures cache is shared across all tool subclasses
- Maintains identical functionality with improved performance
This significantly reduces startup time and resource usage when OpenRouter
is configured, especially noticeable with many custom models.
🤖 Generated with Claude Code
Co-Authored-By: Claude <noreply@anthropic.com>
- Properly mock Redis client operations to support add_turn functionality
- Set up initial thread contexts so add_turn can find existing threads
- Mock Redis set operations to return success
- Ensure all Redis-dependent tests use proper mock patterns
- All 473 unit tests now pass 100% with proper isolation
🤖 Generated with [Claude Code](https://claude.ai/code)
Co-Authored-By: Claude <noreply@anthropic.com>
Fixed NameError that was causing Docker container crashes:
- Updated type annotation in format_response method from TestGenRequest to TestGenerationRequest
- This was the last missing reference from the class rename
🤖 Generated with [Claude Code](https://claude.ai/code)
Co-Authored-By: Claude <noreply@anthropic.com>
Fixed MagicMock comparison errors across multiple test suites by:
- Adding proper ModelCapabilities mocks with real values instead of MagicMock objects
- Updating test_auto_mode.py with correct provider mocking for model availability tests
- Updating test_thinking_modes.py with proper capabilities mocking in all thinking mode tests
- Updating test_tools.py with proper capabilities mocking for CodeReview and Analyze tools
- Fixing test_large_prompt_handling.py by adding proper provider mocking to prevent errors before large prompt detection
Fixed pytest collection warnings by:
- Renaming TestGenRequest to TestGenerationRequest to avoid pytest collecting it as a test class
- Renaming TestGenTool to TestGenerationTool to avoid pytest collecting it as a test class
- Updated all imports and references across server.py, tools/__init__.py, and test files
All 459 tests now pass without warnings or MagicMock comparison errors.
🤖 Generated with [Claude Code](https://claude.ai/code)
Co-Authored-By: Claude <noreply@anthropic.com>
- Exit early at MCP boundary if files won't fit within given context of chosen model
- Encourage claude to re-run with better context
- Check file sizes before embedding
- Drop files from older conversations when building continuations and give priority to newer files
- List and mention excluded files to Claude on return
- Improved tests
- Improved precommit prompt
- Added a new Low severity to precommit
- Improved documentation of file embedding strategy
- Refactor
When directories were provided to tools, only the directory path was stored
in conversation history instead of the individual expanded files. This caused
file filtering to incorrectly skip files in continued conversations.
Changes:
- Modified _prepare_file_content_for_prompt to return (content, processed_files)
- Updated all tools to track actually processed files for conversation memory
- Ensures directories are tracked as their expanded individual files
Fixes issue where Swift directory with 46 files was not properly embedded
in conversation continuations.
When directories were provided to tools, only the directory path was stored
in conversation history instead of the individual expanded files. This caused
file filtering to incorrectly skip files in continued conversations.
Changes:
- Modified _prepare_file_content_for_prompt to return (content, processed_files)
- Updated all tools to track actually processed files for conversation memory
- Ensures directories are tracked as their expanded individual files
Fixes issue where Swift directory with 46 files was not properly embedded
in conversation continuations.
New test to confirm history build-up and system prompt does not affect prompt size checks
Also check for large prompts in focus_on
Fixed .env.example incorrectly did not comment out CUSTOM_API causing the run-server script to think at least one key exists
Improved prompt for immediate action
Additional logging of tool names
Updated documentation
Context aware decomposition system prompt
New script to run code quality checks
Docs added to show how a new tool is created
All tools should add numbers to code for models to be able to reference if needed
Enabled line numbering for code for all tools to use
Additional tests to validate line numbering is not added to git diffs
Supports decomposing large components and files, finding codesmells, finding modernizing opportunities as well as code organization opportunities. Fix this mega-classes today!
Line numbers added to embedded code for better references from model -> claude
Improved testgen; encourages follow-ups with less work in between and less token generation to avoid surpassing the 25K barrier
Improved coderevew tool to request a focused code review instead where a single-pass code review is too large or complex
OPENROUTER_ALLOWED_MODELS environment variable support to further limit the models to allow from within Claude. This will put a limit on top of even the ones listed in custom_models.json
Encourage Claude to pick the best model for the job automatically in auto mode
Lots of new tests to ensure automatic model picking works reliably based on user preference or when a matching model is not found or ambiguous
Improved error reporting when bogus model is requested and is not configured or available
* Support for Custom URLs and custom models, including locally hosted models such as ollama
* Support for native + openrouter + local models (i.e. dozens of models) means you can start delegating sub-tasks to particular models or work to local models such as localizations or other boring work etc.
* Several tests added
* precommit to also include untracked (new) files
* Logfile auto rollover
* Improved logging
