github-profile-optimization

Run the bundled profile-audit.py script with your username to collect facts from GitHub's REST and GraphQL APIs, then score the page against four blocking checks and eight scored checks worth 16 points. Ship when all blocking checks pass and the score reaches 13 of 16.

A profile README renders only when four prerequisites hold at once: the repository name matches your username exactly, the repository is public, README.md sits at the repository root, and the file is not empty. Failures are silent with no error message.

Choose pins against your stated goal, not star count: flagship first for adoption, curated good-first-issues first for contributors, or judgement-demonstrating work for hiring. Never pin archived repositories or trivially modified forks, and give each pin a description and topics first.

The audit script calls GitHub's own APIs and has no equivalent for GitLab, Codeberg, or self-hosted forges like Forgejo. On other hosts, apply the same scorecard checklist by hand from the profile page and state which facts could not be verified.

No published study measures whether profile changes affect interviews, stars, or signups, so the Skill refuses to promise those outcomes. It delivers a page that stops contradicting your other material and routes visitors deliberately.

Only commits on the default branch with an account-linked email count, and commits in forks never count until merged upstream. Enable the anonymized private-contribution setting first, then fix mechanical causes like unlinked commit emails.