Universal Sub-Agent Console

Agent roster

Import / Export

Import roster JSON (replaces current roster — export first if unsure)

Export = one .json file for the whole roster, plus a button that downloads one .md per agent (your browser may ask to allow multiple downloads). See AGENT-FORMAT.md for the portable format.

Agent

Name Role (one line) Tags (comma-separated) Charter (Markdown) — this IS the agent. Paste it into any platform as the system / custom instructions.

▶ Run brief generator

Pick an agent, type the task. The app composes a complete, self-contained brief (charter + task + output format). Copy it into any AI platform, or download it as a .md file.

Agent Task Output format

Composed brief

⚡ Run it here — v2 execution

Runs the agent selected above, with the task and output format from the Run brief section, in this page. Uses your own API key — get a free one from Groq or OpenRouter (see README.md), or use any OpenAI-compatible endpoint, including a local model on your own network.

Provider

API key for this provider

Remember on this device (off = key is forgotten when this tab closes)

Your key is stored only in this browser (session storage by default, local storage only if you tick Remember). It is sent only to the provider you picked above, as that provider's sign-in — never anywhere else, and never into an agent charter, a brief, or an export.

Endpoint base URL Model

Model IDs change over time — the suggestions are starting points; type any model ID the provider offers.

Result

No run yet. Pick an agent and a task above, then RUN.

Follow-up (continues this agent's conversation)