chore: bump project version to 1.52.0 and reformat multi-line strings across core modules
- Update pyproject.toml version from 1.51.0 to 1.52.0 - Reformat string quotes from single to double in autonomous mode extraction - Reformat tool_results list construction and __all__ exports with consistent indentation - Remove unused imports (time, uuid) and Spinner from assistant module - Reformat knowledge_context search results and graph_memory dataclass definitions - Fix missing space in KnowledgeEntry __str__ method's default parameter
This commit is contained in:
@@ -1,6 +1,53 @@
|
||||
# Changelog
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
## Version 1.52.0 - 2025-11-10
|
||||
|
||||
This release updates the project version to 1.52.0. No new features or changes are introduced for users or developers.
|
||||
|
||||
**Changes:** 1 files, 2 lines
|
||||
**Languages:** TOML (2 lines)
|
||||
|
||||
## Version 1.51.0 - 2025-11-10
|
||||
|
||||
The system can now extract and clean reasoning steps during task completion. Autonomous mode has been updated to recognize these reasoning steps and task completion markers, improving overall performance.
|
||||
|
||||
**Changes:** 5 files, 65 lines
|
||||
**Languages:** Python (63 lines), TOML (2 lines)
|
||||
|
||||
## Version 1.50.0 - 2025-11-09
|
||||
|
||||
### Added
|
||||
- **LLM Reasoning Display**: The assistant now displays its reasoning process before each response
|
||||
- Added `REASONING:` prefix instruction in system prompt
|
||||
- Reasoning is extracted and displayed with a blue thought bubble icon
|
||||
- Provides transparency into the assistant's decision-making process
|
||||
|
||||
- **Task Completion Marker**: Implemented `[TASK_COMPLETE]` marker for explicit task completion signaling
|
||||
- LLM can now mark tasks as complete with a special marker
|
||||
- Marker is stripped from user-facing output
|
||||
- Autonomous mode detection recognizes the marker for faster completion
|
||||
- Reduces unnecessary iterations when tasks are finished
|
||||
|
||||
### Changed
|
||||
- Updated system prompt in `context.py` to include response format instructions
|
||||
- Enhanced `process_response_autonomous()` to extract and display reasoning
|
||||
- Modified `is_task_complete()` to recognize `[TASK_COMPLETE]` marker
|
||||
- Both autonomous and regular modes now support reasoning display
|
||||
|
||||
**Changes:** 3 files, 52 lines
|
||||
**Languages:** Python (52 lines)
|
||||
|
||||
## Version 1.49.0 - 2025-11-09
|
||||
|
||||
Autonomous mode is now enabled by default, improving performance. Identical messages are now removed in autonomous mode to prevent redundancy.
|
||||
|
||||
**Changes:** 3 files, 28 lines
|
||||
**Languages:** Markdown (18 lines), Python (8 lines), TOML (2 lines)
|
||||
|
||||
## Version 1.48.1 - 2025-11-09
|
||||
|
||||
### Fixed
|
||||
|
||||
Reference in New Issue
Block a user