Skip to content
EngineeringOpenAlt AI Labs

Ship a real pull request with Claude Code

Claude path, stage 5. Use Claude Code to plan, build and test a real change in an open-source project or a project with real users, and open a pull request for it.

  1. 1Learn
  2. 2Do
  3. 3Submit
  4. 4Results
Category
Engineering
Lesson
Self-paced
Effort
~6h
Prize
Free
Closes in
 

Learn

Take the course first

This stage starts with Claude Code 101, a free course by Anthropic on the Anthropic Academy site. When you are comfortable, Claude Code in Action covers longer sessions: steering, configuring, automating and verifying. Both are in English and self-paced.

Anthropic Academy is made and run by Anthropic. Alt AI Labs is not affiliated with Anthropic: we link to the course and set this challenge.

Work in this order

  1. Read the issue and the contribution guide yourself. Some projects do not accept AI-assisted pull requests; pick another project if so. Skip issues someone else is assigned to, and comment on the issue to ask a maintainer before you start.
  2. Ask for a plan before any code. Read it and correct it.
  3. Let it build, then review every line. You are the author of the pull request, not Claude.
  4. Prove it works. Run the tests, add a test for your change, and try it by hand.
  5. Keep the diff small. Revert anything the issue did not ask for.

Write down what you corrected

The write-up is the part judges read first. One honest paragraph on what Claude got wrong and how you caught it is worth more than a long summary.

Finish the lesson to unlock submitting.

Sign in to start learning

Do

about 6h of work

The task

Claude Code can read a whole codebase, make changes across files and run the tests. The skill is steering it: giving it the right context, checking its plan, reviewing what it wrote and proving the change works. Pick an open issue in an open-source project, such as one labelled "good first issue", or a real feature in a project people use. Check that the project accepts AI-assisted contributions, that nobody else is assigned to or working on the issue, and comment on the issue to ask a maintainer before you start. Open one pull request only. Use Claude Code to plan the change, build it and test it, and open a pull request. Then write a short account of how you worked: the plan, what you changed after reviewing Claude's code, and how you know it works.

Deliverables

  1. 01A link to the pull request
  2. 02A short write-up in the pull request or a linked doc: the plan, what you corrected, and how you verified the change
  3. 03The CLAUDE.md file or project notes you gave Claude Code, if you wrote any

Acceptance criteria

Judged against exactly these

  • The pull request solves a real issue or adds a real feature, and links to it
  • A maintainer agreed on the issue before the pull request was opened, and nobody else was assigned to it
  • The project's contribution rules allow AI-assisted work, and the pull request says Claude Code was used
  • Automated checks pass, and the change includes tests or explains why none apply
  • The write-up names at least one thing you changed after reviewing Claude's code
  • The diff contains only changes needed for the issue
  • One pull request per entrant, with no secrets or personal data in it

Rules

  • Complete the lesson first. The submit form unlocks once you mark it done.
  • Submit a link to your work: a public GitHub repo or PR for code challenges; YouTube, Google Drive or Docs, Figma, Notion or a live site for everything else.
  • AI tools are allowed and encouraged. Plagiarised or copied work is disqualified.
  • The work must be made during the challenge window. For code, the commit history must show it.
  • An AI reviewer pre-scores every entry against the acceptance criteria. A human makes the final decision.
  • On payout, Alt AI Labs receives a licence to use the winning work (MIT for code). Non-winning entries stay yours.
  • Do not include secrets, API keys or anyone's personal data.
claude-codegittestingcode-review

Submit

Sign in and complete the lesson to submit. Entry is free.

Sign in

Results

  1. 01

    AI pre-score

    Every entry is scored by an AI reviewer against the acceptance criteria.

  2. 02

    Human decision

    A person on the challenge team reviews the shortlist and makes the final call. The AI score is advisory.

  3. 03

    Win

    There is no cash prize. Winning work is shown to the organization behind the challenge.

Posted by

Alt AI Labs