Autonomous agents, from the terminal.

Most agent tools put you in the driver's seat — approve this tool call, review this diff, confirm this action. That's useful for interactive work. But there's a whole class of tasks where you just want the work done: morning briefs, market monitoring, PR reviews, research digests, security scans.

Aeon Terminal is built for that. Today, that means a web terminal you sign into, a catalog of skills you can run by name, and a Claude backend with live fetch + RSS tools. Tomorrow, that means cron, fan-out, and self-healing — see the status panel below.

The terminal isn't a theme, it's the surface area. A skill is a command. A run is a stream of coloured lines you can read in three seconds. There's no chat, no thread, no "assistant" with a name. You pick a skill, the skill picks the prompt.

What we're not.

• Not a coding agent. Use Claude Code or Cursor for that.
• Not a chat assistant. We don't want a relationship with your agent.
• Not a no-code builder. Skills are prompts, not workflows you drag and drop.
• Not a yet-another-LLM-frontend. We default to live data and call out hallucinations as bugs.

What today looks like.