Quickstart: your whole agent team, one memory, 5 minutes

Based on the open-source client kit at github.com/LogicMem/LogicMem-mcp- (MIT).

1. Get an API key

Keys are Bearer tokens that start with lm_. Sign up here and copy your key from the dashboard. Treat the key like a password and never commit it to git.

2. Connect your first agent over MCP

Any client that speaks streamable-HTTP MCP can connect. Example for OpenClaw (~/.openclaw/openclaw.json), from the kit README:

{
  "mcp": {
    "servers": {
      "logicmem": {
        "transport": "streamable-http",
        "url": "https://mcp.logicmem.io/mcp",
        "headers": { "Authorization": "Bearer lm_YOUR_API_KEY" }
      }
    }
  }
}

Other clients (Claude, Cursor and similar) use the same URL and header in their own MCP settings. The server also supports legacy SSE.

3. Connect the rest of your agents, from any vendor

Repeat step 2 in every client you use (for example Claude, Cursor and a local model). There is no fixed limit of two: use keys for the same workspace and every connected agent shares the same memory.

4. Run the handoff test

In agent A, ask it to call logicframe_memory_log with: "Handoff test: the deploy freeze starts Friday."
In agent B, ask: "Use logicframe_memory_recall to find anything about a deploy freeze."
Agent B (or any other connected agent) should return the note agent A wrote.
Call logicframe_audit_verify to confirm the audit chain is intact.

Prefer Python?

pip install "logicmem[cli] @ git+https://github.com/LogicMem/LogicMem-mcp-.git"

from logicmem import LogicMem
import os
memory = LogicMem(api_key=os.environ["LOGICMEM_API_KEY"])
memory.log(text="Handoff test: the deploy freeze starts Friday.", category="task", importance=8)
print(memory.recall(query="deploy freeze")[0]["text"])