Execute RunAPI Model Tasks

RunAPI CLI is a skill for using the runapi command-line tool, the execution layer for RunAPI model tasks, to generate AI images, videos, or music/audio, run one-off model jobs, pass JSON request bodies, wait for async tasks, or script RunAPI from a terminal, server, or CI job. It's marked critical risk since commands can call remote services, consume credits, and expose account state.

Installation is via Homebrew on macOS/Linux (brew install runapi-ai/tap/runapi) or, for servers and CI, by downloading an installer script, inspecting it before running, and optionally pinning a specific version - the installer detects OS/architecture, verifies the SHA-256 checksum against a published manifest, and refuses to write the binary if verification fails.

Authentication is treated as security-sensitive throughout: check current auth state with runapi auth status; provide credentials via the RUNAPI_API_KEY environment variable, via --token - stdin import for a saved config, or via interactive browser login only when the user explicitly wants it; RUNAPI_BASE_URL overrides the default endpoint; and secrets should never be passed directly as command arguments.

The CLI is JSON-first and self-documenting - runapi --help, runapi suno --help, and runapi suno text-to-music --help reveal each service's typed commands and request fields before composing a request. Running a model passes a JSON request body via --input-file, --input, or stdin; the default flow is synchronous and polls until completion, or --async submits the task and runapi wait/runapi get check on it later. JSON responses go to stdout and progress lines to stderr, so output pipes cleanly to jq. Account commands (runapi account info, runapi account balance) check status and credit balance. The skill itself can be installed into other agent runtimes (Claude, Codex, Gemini, OpenClaw, or a custom target directory) via runapi agent install-skill.

Use this skill when an agent needs to generate AI images, video, or music/audio via RunAPI, inspect RunAPI auth or account status, submit and wait on async RunAPI tasks, or install/set up the RunAPI CLI on a machine or CI runner.

Do not run interactive runapi login by default from an agent - browser login should only happen when the user explicitly wants it. Do not pass API keys directly in command arguments or paste them into example commands or PR text; always check the CLI exit code before assuming a task succeeded rather than trusting output alone.

Input: a RunAPI service/action name and a JSON request body (file, inline, or stdin), plus valid RunAPI authentication.

Output: a completed model task (image, video, or audio) or an async task handle, run via:

runapi suno text-to-music --input-file request.json

runapi suno text-to-music --async --input-file request.json
runapi wait <task-id> --service suno --action text-to-music

runapi get <task-id> --service suno --action text-to-music

Developers and agent builders who need to generate AI media from a terminal, server, or CI job through RunAPI's typed, JSON-first CLI, with security-conscious credential handling built in.