CLI

rilogtm is Rilo's connector CLI. It detects the AI clients on your machine, writes the Rilo endpoint into each one's MCP config, and tells you how to finish authentication in the client.

Interactive

npx rilogtm connect

"Configuration complete" means the URL was written; it does not mean browser sign-in has finished. Verify the connection from the agent with:

Show my Rilo workspace overview.

Install the operating skill

The MCP connection gives the agent tools. The optional model-invoked skill teaches local agents the canonical staged-search and outbound flow:

npx skills add https://github.com/workatoms/growth-infra-for-agents/tree/main/skills/rilo-gtm --global --agent claude-code --agent codex --agent cursor --agent openclaw --copy --yes

Restart the selected agents after installation. The GitHub URL is the distribution source; each agent discovers the installed SKILL.md from its own local skill directory. Update it later with npx skills update rilo-gtm --global --yes.

Flags

npx rilogtm connect --client claude-code,cursor --yes   # non-interactive
npx rilogtm connect --client all --yes                  # every detected client
npx rilogtm connect --dry-run                           # print plan, change nothing
npx rilogtm connect --url https://other-host/mcp        # override endpoint

The endpoint can also come from the environment:

RILO_MCP_URL=https://other-host/mcp npx rilogtm connect

Headless fallback

For CI or machines without a browser, create an API token from the API tokens section in Setup and use the manual configuration in the client-specific guide. Keep the token in a secret manager rather than shell history.

Scripting against the API

For driving operations themselves from a shell, use the HTTP API with curl - same operations, same envelope.