- Remove database dependency for session management, use chat_id directly
- Add TIMEOUT valve (default 300s)
- Fix streaming issues by handling full message events
- Improve chat_id extraction and tool detection
- Update docs and bump version to 0.1.1
- Add 'Prerequisites' section to folder-memory README files clarifying that conversations must occur inside a folder
- Update docs/plugins/filters/folder-memory.md and folder-memory.zh.md with same prerequisites
- Enhance extract_plugin_versions.py to auto-generate GitHub README URLs in release notes
- Update plugin-development workflow to document README link requirements for publishing
- Rename plugin from `folder-rule-collector` to `folder-memory` for better clarity.
- Refactor code to focus on "Project Rules" collection, removing "Knowledge" collection for V1.
- Add `PRIORITY` valve (default: 20) to ensure execution after context compression.
- Update all parameter names to uppercase for consistency.
- Update documentation (README, global docs) with GitHub raw URL for demo image.
- Remove `STATUS` valve as it's redundant with OpenWebUI's built-in function toggle.
- Add `ROADMAP.md` to track future "Project Knowledge" features.
- Update `.github/copilot-instructions.md` with detailed commit message guidelines.
This release introduces significant improvements to configuration flexibility, performance, and stability.
**Key Changes:**
* **Smart Configuration:**
* Added `summary_model_max_context` to allow independent context limits for the summary model (e.g., using `gemini-flash` with 1M context to summarize `gpt-4` history).
* Implemented auto-detection of base model settings for custom models, ensuring correct threshold application.
* **Performance & Refactoring:**
* Optimized `model_thresholds` parsing with caching to reduce overhead.
* Refactored `inlet` and `outlet` logic to remove redundant code and improve maintainability.
* Replaced all `print` statements with proper `logging` calls for better production monitoring.
* **Bug Fixes & Modernization:**
* Fixed `datetime.utcnow()` deprecation warnings by switching to timezone-aware `datetime.now(timezone.utc)`.
* Corrected type annotations and improved error handling for `JSONResponse` objects from LLM backends.
* Removed hard truncation in summary generation to allow full context usage.
**Files Updated:**
* Plugin source code (English & Chinese)
* Documentation and READMEs
* Version bumped to 1.2.1