@cortexdb/openclaw

Long-term memory plugin for OpenClaw, powered by CortexDB.

Mirrors the same plugin shape as the Mem0 OpenClaw integration: a memory provider with auto-recall, auto-capture, and a set of model-callable memory_* tools.

Install

openclaw plugins install @cortexdb/openclaw
openclaw gateway --force

Configure

In openclaw.json:

Cloud mode (CortexDB Cloud)

{
  plugins: {
    entries: {
      "openclaw-cortexdb": {
        enabled: true,
        config: {
          apiKey: "${CORTEXDB_API_KEY}",
          userId: "alice"
        }
      }
    }
  }
}

Get an API key at cortexdb.ai.

Self-hosted mode

{
  plugins: {
    entries: {
      "openclaw-cortexdb": {
        enabled: true,
        config: {
          mode: "self-hosted",
          apiUrl: "http://localhost:3141",
          userId: "alice"
        }
      }
    }
  }
}

What it does

• Auto-recall. Before every prompt build, the plugin queries CortexDB for memories matching the current message and injects them into the system prompt as a preamble.
• Auto-capture. After every assistant reply, the user/assistant exchange is sent to CortexDB on a non-blocking path.
• Four tools the model can call directly:
  • memory_search — natural-language memory queries
  • memory_list — view recently stored memories
  • memory_store — explicitly save a fact
  • memory_forget — delete by query (requires audit reason)

Why CortexDB

CortexDB is the option you reach for when the builtin SQLite engine and the Honcho plugin stop fitting:

• Distributed, event-sourced, replicated storage
• Hybrid retrieval (vector + keyword + graph + temporal, fused)
• Bitemporal knowledge graph with right-to-be-forgotten and legal holds
• Cross-encoder reranking and adaptive ranker weights
• Multi-tenant isolation with hierarchical scopes

See the docs for the full walkthrough.