Caura — governed shared memory for AI agent fleets

Caura (formerly MemClaw) is an Agent DB — governed shared memory for AI agent fleets. Agents commit what they learn once; every agent in the fleet recalls it through MCP tools, REST, or Caura Rail (preview), subject to tenant isolation, visibility scope (scope_agent / scope_team / scope_org) and caller trust level.

caura-sdk, caura and caura-client all install the same official Python client. caura-client is the canonical name; the other two exist so that install instructions pointing at them resolve instead of 404ing. The import package here is caura_sdk.

Install

pip install caura-sdk

Quickstart

from caura_sdk import Caura

# Get an API key at https://caura.ai, or point base_url at a self-hosted server.
with Caura("mc_xxx", tenant_id="my-team", agent_id="my-agent") as mc:
    # Commit a memory once. The server enriches it with type, title, tags and importance.
    mc.write("Q3 revenue target is $4M, set on 2026-04-15.")

    # Search: ranked raw results.
    for m in mc.search("Q3 revenue target", top_k=5):
        print(m.title, "—", m.content)

    # Recall: an LLM-synthesized context brief.
    print(mc.recall("Q3 revenue target").summary)

Self-hosted? Pass base_url="http://localhost:8000". The full client API is on the caura-client page.

Connect an MCP client

Agents that speak MCP (Claude Code, Cursor, OpenClaw and others) reach the same memory without any SDK. Copy an API key from the caura.ai dashboard and add:

{
  "mcpServers": {
    "caura": {
      "url": "https://caura.ai/mcp",
      "headers": { "X-API-Key": "mc_your_api_key_here" }
    }
  }
}

For a production fleet, provision one agent-scoped credential per agent; see per-agent keys. With a tenant-scoped dashboard key, pass an explicit agent_id on every tool call.

Three ways agents use memory

• Deterministic: Caura Rail (preview). Your code runs recall before every agent turn and commit after it, so the model cannot skip the memory step.
• Agentic: MCP tools or REST. The agent decides when to call caura_write and caura_recall, through the MCP config above or this client.
• Reflective: the Interviewer. Caura reads the agent's own session transcript after the fact and stores the decisions and preferences the agent never stopped to record.

Links

• Documentation: https://caura.ai/docs
• Source: https://github.com/caura-ai/caura (Apache-2.0)
• Issues: https://github.com/caura-ai/caura/issues
• Benchmark: https://github.com/caura-ai/caura-longmemeval (LongMemEval harness)
• Canonical package: caura-client