[WIP Feature] Introduce Knowledge Evolver Module

[fanqiNO1]

🚀 Runtime Knowledge Evolution

• Knowledge Evolution Framework: Introduced a signal-driven KnowledgeEvolver that organizes the runtime knowledge graph after each search — connecting and merging similar clusters, refreshing edges as embeddings drift, detecting communities, and periodically refining low-quality clusters — so the knowledge base improves with use.
• Meta-Cluster Routing: Community detection builds meta clusters as a directory layer over the knowledge graph; similarity retrieval now routes through them, reducing per-query computation while keeping meta clusters out of recall.
• Storage Enhancements: Extended KnowledgeStorage with joint cluster+embedding reads, richer merge semantics (queries, search results, merge counting), and reverse-edge cleanup on deletion.
• Non-Blocking Integration: Evolution runs fire-and-forget after each search behind a kill switch, with reentrancy locking and LLM concurrency control, so search latency is unaffected.

[gemini-code-assist]

Code Review

This pull request introduces a runtime knowledge cluster evolution system via the new KnowledgeEvolver class, which manages connecting, merging, and refreshing knowledge clusters. Key changes include integrating this evolver into the search workflow, updating the storage layer to support cascading edge deletions and deeper cluster merging, and adding leidenalg to requirements. The review feedback highlights critical issues with missing await keywords on several asynchronous database and logging calls, potential runtime TypeErrors during datetime parsing, a missing igraph dependency, and the risk of background tasks being garbage collected prematurely.

Important

The consumer version of Gemini Code Assist on GitHub is being sunset. Starting June 18, 2026, new organization installations will be blocked, and all code review activity will officially cease on July 17, 2026.
For more details on the timeline and next steps, please review the Help Documentation.