Most of my work with agents is conversational — turn by turn, tight collaboration on hard problems. Between turns there's a natural gap. Without Tend, that gap became tab-scanning, terminal-hopping, trying to remember which project was waiting. Every tool I looked at built a dashboard. A dashboard is a permanent invitation to break focus.

What I needed was something that speaks only when spoken to. A dot in my shell prompt (○ / ●2) tells me whether anything needs me after every command. Type td at any natural pause for a 3-second scan. Pull, not push.

Agents emit one-line events to a .tend/events file. Git activity as fallback. No daemon, no database. The relay brings in remote agents — Codex, CI, SSH sessions — onto the same board. AI insights tell you what each agent is actually doing and what comes next.

Happy to answer any questions!

I built Tend so I could maximize the number of projects I work on simultaneously without burning out from context-switching between them, checking who needs me, and keeping them all stoked wherever they are. One board shows every agent across every project: who's working, who's done, who needs me.