Engram Setup

Configure the connection to your Engram server by editing ~/.claude/settings.json.

Instructions

1. Determine the server URL

Ask the user:

What is your Engram server address? (e.g., http://192.168.1.100:37777 or http://engram.local:37777)

If the user is unsure, suggest checking their Docker host's IP and port 37777.

Store the answer as SERVER_URL. The MCP endpoint is {SERVER_URL}/mcp.

2. Determine the API token

Ask the user:

What is your Engram API token? (Leave empty if you haven't set one on the server.)

If empty, note that the server must also have no ENGRAM_API_TOKEN set (or it will reject requests).

Store the answer as API_TOKEN.

3. Update settings.json

Read ~/.claude/settings.json, then add ENGRAM_URL and ENGRAM_API_TOKEN to the env section. Use the Edit tool to modify the file directly.

If the env section already exists, add the two keys to it. If it doesn't exist, create it.

Example result (env section only):

{
  "env": {
    "ENGRAM_URL": "http://192.168.1.100:37777/mcp",
    "ENGRAM_API_TOKEN": "your-token-here"
  }
}

Replace values with the actual answers from steps 1 and 2. If the token is empty, set it to "".

4. Restart Claude Code

Tell the user:

Settings are only read when Claude Code starts. Please close and reopen Claude Code for the changes to take effect.

5. Verify connection

After the user restarts and returns, run the health check:

Tool: check_system_health()

• Success: Report the server version and observation count. Setup is complete.
• Failure: Run /engram:doctor to diagnose the issue.

Common issues

• URL must include /mcp: The correct format is http://host:37777/mcp, not just http://host:37777.
• Token mismatch: The token here must match ENGRAM_API_TOKEN (or API_TOKEN in .env) on the server.
• Firewall: Ensure port 37777 is open between this machine and the server.
• Docker networking: If the server runs in Docker, use the host machine's IP, not localhost (unless running on the same machine).