Skip to content

BrainLayer

Persistent memory for AI agents. Search, think, recall — across every conversation you've ever had.

Your AI agent forgets everything between sessions. Every architecture decision, every debugging session, every preference you've expressed — gone.

BrainLayer fixes this. It's a local-first memory layer that gives any MCP-compatible AI agent the ability to remember, think, and recall across conversations.

Key Features

  • 17 MCP tools — served by BrainBar on /tmp/brainbar.sock; a session boots into a core palette of 5 (brain_search, brain_store, brain_recall, brain_expand, expand_palette) and expand_palette exposes the rest
  • Local-first — SQLite + sqlite-vec, single file, no cloud, no Docker
  • Hybrid search — semantic vectors + keyword, merged with Reciprocal Rank Fusion
  • 15-field enrichment — summary, key facts, tags, importance, intent, entities, sentiment, and more, via Groq/Gemini/MLX/Ollama
  • Multi-source — Claude Code (batch + real-time watcher), Codex CLI, T3 threads, YouTube, manual
  • Works everywhere — Claude Code, Cursor, Zed, VS Code, any MCP client

The 14 old brainlayer_* names are still handled by the Python library handlers under src/brainlayer/mcp/, but BrainBar — the agent transport — does not serve them. Use brain_*.

Quick Example

pip install brainlayer
brainlayer init              # Interactive setup wizard
brainlayer index             # Index your conversations

Add to Claude Code (~/.claude.json): 

{
  "mcpServers": {
    "brainlayer": {
      "command": "brainlayer-mcp-stdio-bridge"
    }
  }
}

BrainBar must be running and owning /tmp/brainbar.sock. The bridge ships with the package and reconnects across BrainBar restarts. If you already have socat, the manual equivalent is {"command": "socat", "args": ["STDIO", "UNIX-CONNECT:/tmp/brainbar.sock"]} — see MCP Config.

Your agent now has persistent memory. Ask it:

  • "What approach did I use for auth last month?" → brain_search
  • "Open this result in full" → brain_expand
  • "What was I working on yesterday?" → brain_recall
  • "Remember this for later" → brain_store

Architecture Overview

graph LR
    A["Claude Code / Cursor / Zed"] -->|MCP| B["BrainBar MCP Server<br/>17 tools"]
    B --> C["Hybrid Search<br/>semantic + keyword (RRF)"]
    C --> D["SQLite + sqlite-vec<br/>single .db file"]

    E["Conversations<br/>Claude Code JSONL / Codex / YouTube"] --> F["Pipeline"]
    F -->|extract → classify → chunk → embed| D
    G["Local LLM<br/>Ollama / MLX"] -->|enrich| D

Next Steps

BrainLayer / VoiceLayer / cmuxLayer / Golems