Review your agent's plan before it touches anything.

Your coding agent writes a markdown plan, sends you a link (phone works too), you approve / edit / reject, and the agent reads your decision back and acts. Like a pull request — but for what your AI agent is about to do, over an app-less link.

See pricing

App-less link

Recipients open a link in any browser. Nothing to install, nothing to embed.

Read-back loop

The agent blocks on your decision and reads structured results back — approve, requested changes, or reject.

Markdown is the medium

The plan is plain markdown. You edit it directly; the agent diffs your changes.

Open core

The CLI and skill are open source. Run it locally on your LAN for free.

How it works

1 · Plan

The agent calls ow review plan.md before executing.

2 · Decide

You open the link, approve / tweak steps / leave a note.

3 · Continue

The agent reads back your decision and proceeds or course-corrects.