faq
Questions, answered plainly.
What does “stet” mean?+
Stet is the proofreader's mark that means “let it stand” — don't change what's right. That is the whole product: keep what works, and stop the change that breaks it.
Which AI tools work with Stet?+
Claude Code and Cursor connect over MCP. Codex, Copilot, Windsurf and Gemini CLI read the AGENTS.md that Stet keeps up to date. For ChatGPT, Claude.ai or any other chat, paste the output of stet context.
Do I need to change my workflow?+
No. You keep your tools and your editor. Before an AI task you set the scope; after it, you run stet guard and stet check. An optional git hook blocks out-of-scope commits.
Does Stet see or store my code?+
No. Stet is a local command-line tool: no account and no telemetry, and your code never leaves your machine. Its only network call is optional: with a Linear API key, it fetches the ticket you name, sending just the ticket ID. The source is open on GitHub.
Is it a replacement for tests or CI?+
No. Stet uses your tests: it records which ones prove that a feature works, which files each task may touch, and runs them after every AI change. You can also run it in CI.
Does it work with Linear?+
Yes. Set LINEAR_API_KEY in your shell and run stet scope set LIN-142 --allow "…". Stet pulls the ticket into the context pack you give your AI, links it in the PR text, and the git hook adds the ticket ID to commit messages.
How do I try it?+
Run npx stet-cli init in your project's top folder. It needs Node 18 or newer. The CLI is free and open source; a hosted Stet Cloud for teams is planned.