Quick Configuration
Choose the setup that matches your environment.
Before You Start
- Confirm you have the account, endpoint, or API key required for Memory / Knowledge Graph.
- Start with minimum scopes and read-only access where possible.
- Keep secrets in environment variables instead of hardcoding them in JSON.
Last verified: 2026-06-13
Environment: Node.js 18+, @modelcontextprotocol/server-memory over stdio (knowledge-graph memory)
Install command:
npx -y @modelcontextprotocol/server-memoryLocal config
JSON Local config json
{
"memory": {
"command": "npx",
"args": ["-y", "@modelcontextprotocol/server-memory"],
"env": {
"MEMORY_FILE_PATH": "${MEMORY_FILE_PATH}"
}
}
} After copying
Check the Memory / Knowledge Graph JSON before running it in OpenClaw, then verify the server with a real prompt.
Verify prompt
Remember that my name is Alex and I prefer TypeScript, then ask what you know about me.
Expected success output
The server should store the entities and recall them in a later turn from its knowledge graph.
Common Pitfalls & Fixes
- ⚠️ Watch out: Explain benefits; show before/after conversations.
- 🔑 Always store API keys in environment variables, never hardcode them in JSON.
- 🛡️ Start with read-only scopes if available to verify connection safely.
Security checklist
- The knowledge graph persists across sessions; do not store secrets or sensitive personal data in it.
- Keep the memory file in a private location, not a shared or committed path.
Common errors
| Error | Cause | Fix |
|---|---|---|
Memory tools missing | The host was not restarted after adding the server. | Reload OpenClaw and confirm the memory tools appear in the tool list. |
EACCES / cannot write memory file | MEMORY_FILE_PATH points to an unwritable location. | Set MEMORY_FILE_PATH to a writable file, or omit it to use the default. |
Memory does not persist | The default file lives in a temp path that resets. | Set MEMORY_FILE_PATH to a stable, private file path. |
Example Prompts
Once connected, try these prompts to test capabilities:
- Break down a complex task into clear steps.
- Compare two approaches and list tradeoffs.
- Draft a decision checklist for a new MCP setup.
Verification Checklist
- Run 1-2 real prompts to confirm Memory / Knowledge Graph returns usable data.
- Check that error messages are clear enough for troubleshooting.
- Document the required scopes, dependencies, and env vars for future reuse.