Tomrade and other agent tools

Tools like Orca and Cursor run your agents and help you review their code. Tomrade keeps track of what the agents should work on, how far each piece got and what changed. They do different jobs, and they work together.

Who does what

Agent tools and Tomrade answer different questions.

Running agents and reviewing their code. How do I start several agents at once, keep them from stepping on each other, and look over what they wrote? That’s where tools like Orca and Cursor focus.
Keeping track of the work. What should the agents work on, in what order, how far has each piece got, what changed, and how do I take a change back? That’s Tomrade’s job.

Tomrade doesn’t run a model of its own for your code, and it isn’t an editor. Your coding agent (Claude Code, Codex, Cursor’s agent, or another) does the coding wherever you run it; Tomrade hands it the ticket and follows the result.

Orca

Orca calls itself “the agent development environment”. It’s a free, open-source (MIT) desktop app for macOS, Windows and Linux that runs Claude Code, Codex and other CLI agents side by side, “each in its own worktree”, with split terminals, a browser, diff annotations sent back to the agent, GitHub and Linear review inside the app, worktrees over SSH, and a phone companion to watch agent status.

Where they meet: an agent running in an Orca worktree is still a Claude Code or Codex session, so it can reach Tomrade over MCP like any other. It claims a ticket, works on the ticket’s branch, and Tomrade moves the ticket as the branch, the pull request and the merge come in.

Cursor

Cursor is an AI code editor. Its cloud agents (formerly background agents) run on remote machines, build features and fix bugs on a branch of their own, and open a pull request when they finish. You start one from the editor, from cursor.com/agents, or by mentioning @Cursor in Slack, GitHub or Linear.

Where they meet: Cursor connects to MCP servers, so the agent in your Cursor editor can read, claim and update Tomrade’s tickets. A cloud agent runs on a remote machine and can’t reach Tomrade’s MCP server, which listens on your computer only; give its branch the ticket’s key (tom-42/…) and Tomrade still follows that branch and its pull request back to the ticket through git.

Using them together

One afternoon, start to finish:

In Tomrade: say what you want. The assistant turns it into tickets sized for one pull request, with the dependencies between them.
In your agent tool: open a worktree in Orca, or a chat in Cursor, and tell the agent to pick up TOM-42 from Tomrade. It reads the ticket over MCP and claims it.
While it works: review its diff where you’re used to, in that tool. The ticket in Tomrade shows the branch, then the pull request.
When it merges: the ticket moves to Done on its own. Anything the agent wrote in Tomrade along the way (a new ticket, an edited note, a changed setting) is in History, grouped under that run.

Prefer to stay in one place? Tomrade can also start the agent itself: press Start on a ticket, or let an auto run work through the dependency graph overnight.

What Tomrade’s undo covers

Undo takes back what Tomrade wrote, or let an agent write, through Tomrade: tickets, files and notes in Files, and settings, grouped by run. It does not reach your repository. Commits, branches and pull requests belong to git and your git host, and you revert them there, with the review and protection those tools already give you. Tomrade doesn’t replace them and doesn’t claim to.

How this page was checked

What this page says about other tools comes from their own websites and documentation, read on the date above, and links to them. Products change quickly; if something here is out of date, tell us on GitHub and we’ll fix it.