Your AI's
memory layer.

RAG memory, semantic cache, and intelligent compression. Local-first. Open source. One MCP endpoint.

Get Started FreeView on GitHub
mcp-config.json
{
  "mcpServers": {
    "context-hub": {
      "url": "http://localhost:3100/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_TOKEN"
      }
    }
  }
}

5 MCP Tools — Everything your AI needs

memory_upsert

Store & embed documents

memory_search

Semantic similarity search

context_pack

Bundle optimized context

context_compress

Compress to save tokens

proxy_call

Call sub-MCP servers