# Jules playbooks

Jules is Google's asynchronous coding agent. It clones a GitHub repository into a cloud virtual machine, writes a plan for you to approve, then makes the change and offers it as a branch or pull request. People use it for bug fixes, tests, docs and small features while they do other work.

- Vendor: Google
- Kind: coding-agent
- Surfaces: web, terminal, api
- Connects: MCP support covers a vetted list of servers (Linear, Stitch, Neon, Tinybird, Context7, Supabase) that you connect with API keys in Settings. Jules reads AGENTS.md in the repo root. The Jules Tools CLI and a REST API start and manage tasks.

- Docs: https://jules.google/docs

## Good at

- Background coding tasks on a GitHub repository while you work on something else
- Showing a plan before it edits code, so you can correct the approach early
- Scheduled and suggested tasks, plus fixing CI or Render build failures on its own pull requests

## Limits

- The docs describe only GitHub as the code source.
- Daily and concurrent task caps depend on your Google AI plan. When you hit the cap, you cannot start new tasks until the rolling 24-hour window frees one.

- Setup scripts cannot run long-lived processes such as dev servers or watch scripts.
- Each task starts in a fresh VM, so a broken setup script or a vague prompt often makes the task fail.

## Playbooks (1)

- [Turn small tracker issues into reviewed pull requests with cloud coding agents](https://agentshandbook.com/p/turn-issues-into-pull-requests-with-cloud-agents.md): Each chosen issue gets a pull request with a passing test run that you review, ask to change, merge or close. You stay in control of every merge.
 (trust: corroborated, risk: medium)

## Lessons

- [Background coding agents open pull requests faster than people can review them](https://agentshandbook.com/lessons/background-agent-prs-overwhelm-reviewers.md)
- [Coding agents change, skip or misreport tests to reach a green run](https://agentshandbook.com/lessons/coding-agents-weaken-tests-to-pass.md)