Understand what
your agents ship
When anyone can run ten agents, the scarce thing is a human who understands what ships.
Run your agents, understand what they wrote, decide what ships. Use Threads to shape a plan, Workflows to run dozens of agents that implement it, and the Review Room to understand what changed before it ships.
brew install --cask pago/simpleedit/simpleedit Also for Windows and Linux. Needs Claude Code, Codex or OpenCode installed.
Running agents is easy now. Reviewing them is not.
Every editor can run many agents in parallel. Someone still has to read what they wrote and stand behind it. Most tools answer with more bots. SimpleEdit helps the human instead.
Two agents per change: one builds, one explains
One agent writes the change. A separate, read-only agent that didn't write it helps you understand it: an overview, a reading order, answers to your questions. It never posts, approves or merges. Nothing goes to GitHub that you did not write or accept.
A stack is one review
A chain of dependent PRs opens as one review, layer by layer. Each layer shows only its own changes.
Fits your team as it is
Review a colleague's PR or stack with the same overview, reading order and companion you use for your agents' work. Your review goes out through GitHub, in your name. Nobody else needs to install anything.
Native agents, not a wrapper
Claude Code, Codex and OpenCode run in their own unmodified terminal UI. You keep your login, your subscription, your settings, your MCP servers and your slash commands.
Shape the plan with Threads
Ask the agent for a plan doc. Read it rendered, and comment on any paragraph. The agent answers in the thread and changes the plan. A mistake in a plan costs one comment, not a review round.
- Comment on rendered markdown, the editor or a diff
- The agent answers when it is idle, never mid-turn
- Threads move with their lines when the plan changes
Run large projects with Workflows
Large projects without context bloat. A lead agent turns your plan into a task graph that hidden workers carry out, each in a fresh context. The lead wakes only for decisions, so it keeps the project view.
- When a worker fails, fix the plan, patch the running graph and go on
- Review gates stop the run until you pass them
- Open any worker's terminal and talk to it yourself
Review with a sparring partner, not a bot
Other tools pile comments on your PR. In the Review Room, an agent that did not write the change talks it through with you: how it fits into the system, what it changes, where it could break. You come out understanding the change, ready to approve it or with one clear suggestion you stand behind.
- Ask how it fits in. It reads the code at the PR head.
- It drafts the comment from your talk. You choose what to keep.
- You decide. It never approves or posts.
Keep building and reviewing when you leave your desk
Your agents keep running on your Mac. From your phone you see which one needs you, answer it, start the next session from your backlog, and work through a review with the companion. Threads and drafts are the same ones you see on the desktop.
- A push notification when an agent or a review gate needs you
- The full Review Room: threads, companion and drafts
- Reaches your Mac over Tailscale or OpenTunnel
Your review queue, triaged
Screen PRs reads every PR that waits for you and sorts the queue. Open any PR, or a pasted link, as a stack in the Review Room.
Screen PRs
Triage your review queue into "Needs your attention", "Quick pass" and "Waiting on author". Screen on a schedule, and pick up where you left off with "Continue reviewing".
Sessions
Durable agent sessions, each with its own editor, diff and git log. Group, fork, hand off to a fresh context, or resume after a restart.
Backlog
Sessions you have prepared but can't start yet, because they would collide with running work. A running agent can park a follow-up there and stay on its task. Start it when the coast is clear.
Phone companion
Sessions, backlog, Screen PRs and the full Review Room on your phone, over Tailscale or OpenTunnel, with push notifications.
Providers and models
Claude Code, Codex and OpenCode, plus local models through Ollama. Pick the model per session, and see the tokens each session used.
Your agent can drive it
Every session gets SimpleEdit's tools. Ask your agent to show you a diff, start or park a session, message another agent, ask you for a review or run a workflow.
Claude memory, in view
See what Claude Code remembers about your project, with broken links and unindexed files flagged. Every session of a project shares one memory. Fix it like any other file.
A real editor
Monaco, the editor of VS Code, with language servers for TypeScript, CSS, JSON, Rust, Python and Go.
Try SimpleEdit
Alpha software: expect rough edges. We build SimpleEdit with SimpleEdit, so the sharpest edges get fixed fast.