Best matches for Twill
Twill works best when it can run your full development environment and verify changes mechanically:- Monorepos: A single sandbox run can bring up your whole repo (multiple packages + services) and execute repo-wide checks.
- Web apps: Twill can smoke test UI changes with
agent-browser(including screenshots) alongside your normal build/test/lint steps. - Desktop apps: For Tauri, Electron, and React Native, Twill can run smoke checks via computer-use style UI interaction (click/type/screenshot) when the app is runnable in the sandbox.
What happens when you create a task
At a high level, every task follows the same shape:- You describe what should change (from GitHub, Slack, Linear, Notion, or the web app)
- Twill asks questions when needed (or you can request a plan)
- Implementation happens in a sandbox environment
- Twill runs verification (build/tests/lint when applicable)
- Twill opens a pull request with a summary and artifacts (logs, screenshots, previews depending on your repo)
What you get back
- A pull request with proposed changes
- Verification output (when applicable for your repo)
- Proof-of-work artifacts when available (for example, screenshots for UI changes)
How you stay in control
- Work is delivered as a pull request for human review (nothing merges automatically)
- Code runs in isolated sandboxes, not on your infrastructure
- Verification runs mechanically (tests/build/lint) and results are attached to the PR when available