github-reference-research

Define the feature and the dimensions that matter (implementation, architecture, database, API, framework), search for candidate repositories, triage them by quality criteria, read only the key files, and extract the decisions and trade-offs rather than copying code.

Evaluate seven criteria: recent activity, code quality, test coverage of critical cases, documentation, real adoption beyond stars, license compatibility, and clean architecture. Discard repositories that fail these checks before reading their implementation.

Only if the license permits it; MIT and Apache licenses are generally permissive while GPL and copyleft licenses impose restrictions. This Skill emphasizes extracting principles and patterns rather than copying code, and adapting them to your own context.

Stars measure popularity, not engineering quality; a popular repository can have poor architecture or be unmaintained. Always verify recent activity, tests, and the actual implementation before trusting a repository as a reference.

Use GitHub research when official documentation does not cover real-world implementation trade-offs, such as architecture decisions, data modeling, or error handling patterns. Combine it with official docs and real products rather than relying on a single repository.