Ebi Agent Chat Relay

Ebi Agent Chat Relay

One cockpit for all your coding agents. Run Claude Code, OpenAI Codex, local models, pi and
AG-UI agents side by side — and talk to them wherever you like: a dedicated web console on your
desktop or phone (with Entra ID sign-in), Discord, or Microsoft Teams. Let a strong model manage,
hand the legwork to cheaper ones, and step in only when a decision is actually yours.

日本語 | 简体中文 | 한국어 | Español | Português | Français

Every screenshot in this README uses fictional sample data. Regenerate them with
scripts/readme_screenshots.cjs.

Why people use it

🖥️ The Ebi AI Console — supervise agents, don't babysit terminals

Running several agents is easy; keeping track of them is the hard part. The console answers one
question first: "what do I do now?"

- Focus view — a cheap, tool-less model reads each thread with new activity and fills in
  priority, due date, the next step and whether it needs a particular place (📍 office, home).
  Your next move is pinned at the top. Anything a person sets by hand always wins.
- Focus modes — 🎯 one thread full screen, or 🔁 work through everything that needs you one
  thread at a time: send, and the next one opens. An estimate of your time per item (⚡ quick /
  ⏱ short / 🧘 deep) lets you clear only the quick ones when you have five minutes.
- Sorted by whose move it is — ❓ waiting for your answer, 👀 to review, 📋 your task,
  ⚠️ failed, running, idle, done. Agents report outcomes through the relay, so the board stays
  accurate without extra prompting.
- Act in place — reply with files attached; preview images, Markdown, HTML, CSV and logs inline;
  autocomplete slash commands and skills; ⏸ pause, ■ stop and ▶ resume running work.
- See the cost before it bites — 5-hour and weekly quota meters for Claude and…