Traction scoreGitHub stars can be faked, so popularity alone can be misleading. Traction Score looks for broader signs of recent attention, adoption, and active maintenance.
This library provides a set of structured skills designed to align AI agents with specific engineering standards. It helps platform and developer experience teams maintain consistency across projects by automating the creation of agent instruction files and enforcing commit conventions.
By implementing these skills, teams can ensure that coding agents follow project-specific conventions and that version control history remains clean and reference-backed.
WHO IT'S FOR
Platform / DevEx teams
standardizing AI agent instructions and permissions
Technical writers
drafting engineering blog posts and deep-dives
Product designers / UX writers
creating user-facing UI copy and error messages
Backend / infrastructure engineers
configuring AI coding assistant permissions for repos
Compatible AgentsThe repository documents support for these agents. The skills may also work with other agents that can load SKILL.md files, but they may need some setup or small changes.
<repository-root>: path to the repository whose agent instructions should be drafted
Using the agents-md skill, inspect the repository at <repository-root> and draft the smallest useful root AGENTS.md. Verify its paths and commands against the repository, reference existing documentation instead of copying it, and keep the draft under 60 lines when practical.
<diff-or-PR>: the code diff, pull request, or changed files to review
Relevant project requirements or architecture context, if available
Using the code-review skill, review <diff-or-PR> for runtime errors, performance problems, unintended side effects, compatibility risks, security vulnerabilities, and missing tests. Return concise, actionable findings and identify any change that warrants senior-engineer review. Do not modify the code.
<repository-root>: path to the repository containing the change
<issue-reference>: optional issue identifier to close or reference
Using the commit skill, inspect the coherent change in <repository-root> and draft a Sentry-style conventional commit message without creating the commit. Include an optional body only when useful, and use Fixes or Refs with <issue-reference> only if one is supplied.