mirror of
https://github.com/catlog22/Claude-Code-Workflow.git
synced 2026-02-04 01:40:45 +08:00
4061ae48c4
- Added integration tests for adaptive RRF weights in hybrid search. - Enhanced query intent detection with new classifications: keyword, semantic, and mixed. - Introduced symbol boosting in search results based on explicit symbol matches. - Implemented embedding-based reranking with configurable options. - Added global symbol index for efficient symbol lookups across projects. - Improved file deletion handling on Windows to avoid permission errors. - Updated chunk configuration to increase overlap for better context. - Modified package.json test script to target specific test files. - Created comprehensive writing style guidelines for documentation. - Added TypeScript tests for query intent detection and adaptive weights. - Established performance benchmarks for global symbol indexing.
1.9 KiB
1.9 KiB
CodexLens – Optimization Plan Changelog
This changelog tracks the CodexLens optimization plan milestones (not the Python package version in pyproject.toml).
v1.0 (Optimization) – 2025-12-26
Optimizations
-
P0: Context-aware hybrid chunking
- Docstrings are extracted into dedicated chunks and excluded from code chunks.
- Docstring chunks include
parent_symbolmetadata when the docstring belongs to a function/class/method. - Sliding-window chunk boundaries are deterministic for identical input.
-
P1: Adaptive RRF weights (QueryIntent)
- Query intent is classified as
keyword/semantic/mixed. - RRF weights adapt to intent:
keyword: exact-heavy (favors lexical matches)semantic: vector-heavy (favors semantic matches)mixed: keeps base/default weights
- Query intent is classified as
-
P2: Symbol boost
- Fused results with an explicit symbol match (
symbol_name) receive a multiplicative boost (default1.5x).
- Fused results with an explicit symbol match (
-
P2: Embedding-based re-ranking (optional)
- A second-stage ranker can reorder top results by semantic similarity.
- Re-ranking runs only when
Config.enable_reranking=True.
-
P3: Global symbol index (incremental + fast path)
GlobalSymbolIndexstores project-wide symbols in one SQLite DB for fast symbol lookups.ChainSearchEngine.search_symbols()uses the global index fast path when enabled.
Migration Notes
- Reindexing (recommended): deterministic chunking and docstring metadata affect stored chunks. For best results, regenerate indexes/embeddings after upgrading:
- Rebuild indexes and/or re-run embedding generation for existing projects.
- New config flags:
Config.enable_reranking(defaultFalse)Config.reranking_top_k(default50)Config.symbol_boost_factor(default1.5)Config.global_symbol_index_enabled(defaultTrue)
- Breaking changes: none (behavioral improvements only).