Agents Studio

A coding agent manager runs Claude Code tasks on schedules or pull request events, checks work with your tests, and provides an editor, shell and app preview.

Agents Studio manages Claude Code work on your repositories for developers who want tasks to run while they're away. It brings unattended runs and the tools to review and finish their changes into one interface.

Scheduled and event-driven work: Tasks can start on a schedule or when a pull request opens, then check their changes against your project's tests.
Failure handling: Scheduled work retries a genuine failure once and pauses after repeated unsuccessful runs, with an explanation; tool refusals don't trigger a retry.
Review and editing: You can see summaries of changes, costs and decisions that need your attention, then edit files, use a shell and preview the app in the same interface.
Context and project rules: New runs receive a standing brief assembled from local work records and facts you write, while proposed lessons become file diffs you can accept or decline.

Agents Studio reads your existing .claude directory and the repositories you select. The work it displays and writes uses ordinary files and Git branches on your machine. It's open source under the MIT license.

An optional GitHub issue integration posts one comment when a session started from an issue opens a pull request. The comment includes a summary, the pull request link and a notice that the work hasn't been reviewed. This feature is off by default and doesn't edit, close, reassign or relabel issues.