Runs screen
See every AI agent running on your computer in one place, with what needs you first.
How do I see all my AI agents at once? Open the Runs screen with Ctrl+Shift+M (⌃⇧M on a Mac), or the grid icon in the titlebar. It lists every agent run on your computer in one place, and puts the ones that need you first.
The three groups
| Group | What's in it |
|---|---|
| Needs you | An agent stopped on an approval or a question, and a CLI coding agent (Claude Code, Codex and others) that went quiet waiting for you |
| Running | An agent in the middle of a turn |
| Recent | Your saved conversations, newest first |
Each group shows a count, and an empty group is hidden. The pill in the header says how many runs need you ("1 needs you", "3 need you"), or All runs when none do.
Each active row says where it runs, for example Local · api for an agent in your tab named "api".
What you can do
- Go to a run. Click a row under Needs you or Running to jump to its tab and pane.
- Reopen a saved run. Click a row under Recent to open that conversation in the active pane.
- Search. The Search box filters every group by title.
- Filter. The dropdown narrows the list: Local shows only runs on this computer, and the source options show saved conversations by where they came from (a plain conversation, an orchestration lead or one of its workers).
- Start one. New agent opens a new tab with a fresh conversation.
- Close. The ✕ or Esc takes you back to the terminal.
What a saved run shows
Each row under Recent shows its age, then its source, folder, model, turns, tokens used and run time. The Runs screen shows tokens and time, not credits; what an answer cost is on the usage line under it in the conversation, and in Billing and Usage.
Steps: run three agents and keep track
- Open a tab per task (Ctrl+Shift+T, ⌘T on a Mac), each in its own folder or worktree.
- Start an agent in each with Ctrl+Enter (⌘Enter on a Mac).
- Press Ctrl+Shift+M (⌃⇧M) whenever you want the overview.
- Answer anything under Needs you first, then let the rest run.
The Runs shortcut keeps working while a full-screen program such as a CLI agent has the keyboard. Mel also tells you through notifications when a run needs you, so you don't have to keep the screen open.
FAQ
Is there an agent management panel in Mel? Yes, the Runs screen is it: one list of every agent run on your computer, with what needs you at the top.
Do Claude Code and Codex show up? Yes, when they run in a Mel tab. A CLI that goes quiet waiting for you appears under Needs you.
Where are older conversations? Under Recent, and in the conversations panel (Ctrl+Shift+Y, ⌃⇧Y on a Mac). See Conversation management.