Skip to main content
The agent-cards CLI does everything the dashboard does, from the terminal. Personal commands are top level. Organizations use the companies namespace.

Install and sign in

A sign-in link and short code appear, and you approve in your browser. For scripts and CI, agent-cards login --email you@example.com sends a one-time code and --code 123456 completes it without a browser. Update any time with agent-cards update. Then pick the organization you work in:

Organization commands

All of these live under agent-cards companies. Sandbox or production follows the credentials in use, the same rule as the API. Use --help on any command for flags:

Personal commands

The same CLI runs a personal Agentcard account. Useful for testing what your users will experience.

For coding agents

Two commands turn the CLI into a tool surface an agent can drive without prompts. agent-cards api is the tool catalog: every account tool, with JSON on stdout and {error, hint} envelopes on failure.
agent-cards agents add teaches Claude Code, Codex, or Gemini to drive Agentcard by installing steering blocks that refresh themselves when the CLI updates. agent-cards setup-mcp configures the MCP server in Claude Code in one step. Most interactive commands also accept --yes to skip confirmations and --json for machine-readable output, for example cards create --amount 25 --yes --json.

Rules for your agent