# Devin playbooks

Devin is an autonomous software engineering agent from Cognition. Each session gets its own workspace with a shell, an editor and a browser. Teams hand it tickets from Slack, Microsoft Teams, Linear or Jira, and it returns pull requests.

- Vendor: Cognition
- Kind: coding-agent
- Surfaces: web, slack, terminal, api
- Connects: Install MCP plugins from a marketplace or add custom STDIO, SSE or HTTP servers under Customize. Skills are SKILL.md files in your repos. Knowledge, Playbooks and AGENTS.md give it context. Devin for Terminal is a local CLI.

- Docs: https://docs.devin.ai/

## Good at

- Well-scoped tickets with a clear end, such as passing tests or green CI
- Reproducing a bug and checking a fix by running the app in its browser
- Migrations, refactors, unit tests and PR review across a team's repositories

## Limits

- Cognition's guidance says tasks that would take a person about three hours or less work best. Split bigger projects into smaller sessions.

- Vague requests and a poorly set up environment lead to weaker results.
- Usage is metered in ACUs, and admins can set organization and per-user ACU limits.
- It can repeat the same mistake in later sessions unless you save Knowledge, Skills or corrections.

## Playbooks (2)

- [Map an unfamiliar codebase into a checked architecture document before you change it](https://agentshandbook.com/p/map-an-unfamiliar-codebase-before-you-change-it.md): You have a docs/architecture.md that names the entry points, modules and data flow with file references, a list of open questions, and a short instruction file with the build and test commands.
 (trust: corroborated, risk: low)
- [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[Coding agents write code for old library versions or invent APIs](https://agentshandbook.com/lessons/coding-agents-use-outdated-library-apis.md)
- [Coding agents change, skip or misreport tests to reach a green run](https://agentshandbook.com/lessons/coding-agents-weaken-tests-to-pass.md)