Romp

AI agents like Claude Code can work autonomously for long stretches, allowing several to be run in parallel. But this parallelism creates new management work: tracking what the agents are doing, scrolling through transcripts to find the background a decision needs, and coordinating handoffs of work and context.

Romp simplifies and automates this management: it tracks every agent and its tasks, surfaces what needs your attention, and keeps them moving and working together. *You* stay focused on what you're trying to accomplish.

Every session, one view

The Sessions pane shows every session on a timeline: what's running, what's idle, and what needs you.

Task management

Romp reads the agents' transcripts and tracks the work, so you don't have to.

A card shows each task with the essential information: why the work is happening, what the agent did, and the sub-tasks completed along the way.

Coordination you can see

The Romp Postal Service lets agents ask each other questions and hand off work.

Navigate anywhere, detail on demand

Find any work by when it happened on the timeline or by the task it belonged to, then open it for the full detail.

Every machine, one place

Sessions on your server appear alongside your laptop's, agents hand off work across machines, and you can view everything from a laptop or a phone.

Romp works with Claude Code today. It adds all of this on top of the sessions you already run, without changing how you work.

Self-hosted, reachable from anywhere

You run Romp yourself, on your laptop or a server, with no hosted service in between.

- On your phone. Reach the full dashboard over Tailscale.
- Across machines. Connect several over SSH: agents on different machines message each other, and you steer them all from one place.
- In your editor…