Workshop Recording — Build an AI Recruiter Agent with Trinity

The recording of June 18's workshop is now published. From an empty Claude Code session to a fully deployed, autonomous recruiter — live, in about 38 minutes. No edits, no demo magic. Just the real workflow used to ship agents.

In this workshop we build "Scout," a recruiter agent that defines multiple job specs, interviews candidates over Telegram (voice messages included), scores them against a custom rubric, and hands back a report so a human still makes the hiring call. Then we deploy it to Trinity from a GitHub repo and put a real open role through it.

In this video:

• Scaffolding a new agent from scratch with the create-agent plugin
• The playbook pattern: read state, do the work, save state — and why it keeps agents coherent
• Deploying to Trinity from a GitHub repo (the reliable way) and the local-to-remote sync model
• Wiring an agent to Telegram with email auth + human-in-the-loop approval before access
• A live moment where the agent flat-out refuses to fabricate an answer — and why that is a feature, not a bug

If you have been running agents on your laptop, this is the gap: an entire department deployed in the cloud, governed, and synchronized with version control. Clone Trinity and start playing with it yourself — it's free and open source.

Watch it here → youtube.com/watch?v=K7hFWyFIf-Y

Free agent builder workshops every Thursday — ability.ai/workshops