BrainLayer — Knowledge Pipeline
External repo: github.com/EtanHey/brainlayer Formerly "Zikaron". CLI:
brainlayer. MCP tools:brain_search,brain_store,brain_recall.
BrainLayer indexes Claude Code conversation history into a searchable vector database. Query past solutions, code patterns, and debugging sessions.
MCP Tools (3 available)
| Tool | What It Does |
|---|---|
brain_search | Semantic search across past conversations and knowledge |
brain_store | Persistently store memories (auto-type, auto-importance) |
brain_recall | Proactive context retrieval by file, topic, or current state |
Old brainlayer_* names still work as backward-compat aliases.
CLI Quick Reference
bash1# Search past conversations 2brainlayer search "how did I implement authentication" 3brainlayer search "error handling" --project golems 4 5# Index conversations 6brainlayer index 7brainlayer index -p project-name 8 9# Run enrichment (local LLM metadata extraction) 10brainlayer enrich 11 12# Stats 13brainlayer stats 14 15# Interactive dashboard 16brainlayer dashboard
MCP Config
json1{ 2 "mcpServers": { 3 "brainlayer": { 4 "command": "brainlayer-mcp" 5 } 6 } 7}
Storage
text1~/.local/share/brainlayer/ 2├── brainlayer.db # sqlite-vec database (vectors + metadata, ~1.4GB) 3└── prompts/ # Deduplicated system prompts
When to Use
- Finding past solutions: "How did I handle that API rate limiting before?"
- Debugging patterns: "What error messages have I seen with this library?"
- Code reuse: "Find my previous implementation of pagination"
- File history: "What sessions touched this file?"
- Regression detection: "What changed since this file last worked?"
