Build agent interfaces from your API
Create and deploy a CLI in minutes with Petl
Built by engineers from
Agents think in CLIs.
How it works
API docs to production CLI ready for agents.
Paste your spec
Drop in a docs URL or an OpenAPI file. Studio fetches and parses it in seconds.
Get a CLI
Our engine maps every endpoint to commands designed to make agent use ergonomic.
Install with one line
A GitHub release is created automatically with a one-line install script.
What each CLI ships with
Every command, designed for agents.
FAQ
Common questions
MCPs use an agent loop — one tool call, one round-trip. A CLI lets your agent write a shell script: chain commands with pipes, fan out with xargs, filter with jq, all in one shot. It's a real binary your agent uses at the terminal, the same way a developer would.
No. You point the generator at your OpenAPI spec and it emits a complete Go project — commands, flags, auth, pagination, output formatting — all from the spec. The only thing you configure is the CLI name, env-var prefix, and Go module path.
OpenAPI 3.0 and 3.1. JSON and YAML. The parser resolves $ref chains, handles circular references, and validates the spec before generating anything.
Run mycli --help for a full list of groups and subcommands. For machine-readable introspection, mycli --scheme emits a JSON document covering every command, flag, and expected output — no trial and error.
Customizations live in clicreator.yml — rename commands, hide operations, set an env-var prefix. When your API changes, update the spec, bump the version, and petl publishes a new GitHub release. Your users get the update with a single command.
API key (header and query param) and HTTP Bearer. OAuth flows are planned for v2. Credentials resolve from CLI flags, environment variables, and a config file in that order of precedence.