AgenTrace

Connect your agent

Give your AI agent persistent memory across sessions. Decisions, findings, and context carry forward automatically.

Step 1

Get your credentials

Choose how you want to authenticate. API key is simpler for automation; email and password works if you don't want to manage a key.

Copy your API key from the dashboard. You'll pass it as AGENTRACE_API_KEY in the next step.

Open dashboard

Step 2

Configure your agent

Both agents use @theagentrace/mcp — no separate install needed. Pick your agent then your preferred config method.

bash — API key
claude mcp add \
  --scope user \
  --env AGENTRACE_API_KEY=your_key \
  AgenTrace -- npx -y @theagentrace/mcp
bash — email + password
claude mcp add \
  --scope user \
  --env AGENTRACE_EMAIL=you@example.com \
  --env AGENTRACE_PASSWORD=your_password \
  AgenTrace -- npx -y @theagentrace/mcp

--scope user stores config in ~/.claude.json and works across all projects. Use --scope project to write a .mcp.json shareable with your team — leave credentials out of that file and set them in your shell instead.

Step 3

Verify

Restart your agent after saving config, then confirm the server connected.

bash
claude mcp list
# AgenTrace   connected

Once connected the agent gets session_context, query_memory, save_memory_node, and related tools — no further setup.

Self-hosted

Point to your own server

If you run your own AgenTrace instance, add AGENTRACE_URL. Omit it to use the hosted service.

env
AGENTRACE_URL=https://your-instance.example.com
Ask AI about AgenTrace