Repositories that use the technology (github-repo)

Real projects show how people actually structure code with the technology: which features they use, in which combinations, at what scale. The technology's own repo is covered by package-internals; this hunter looks at everyone else.

Search

• Awesome lists: codelabgen github repos "awesome <topic>". A good list is a source itself (structure), and a map to more repos.
• By name and topic: codelabgen github repos "<topic> in:name,description,readme" --sort stars, or with GitHub topics: topic:<topic>.
• Who depends on it: code search in dependency files, for example codelabgen github code '"langgraph" filename:pyproject.toml' or '"@langchain/langgraph" filename:package.json'. This finds real projects that keyword search misses. Code search allows 10 requests per minute: make each query count.
• Official orgs: the maintainers' other repos (starter templates, example apps): codelabgen github repos "org:<org>".

Check each candidate

codelabgen github meta <repo…> gives stars, last push, license, archived status and topics. Then fetch the README (codelabgen fetch https://github.com/<repo>) and look at the dependency file.

• Recent: pushed within the last year and not archived.
• On the pinned major version: check the dependency file.
• Readable: a clear README and a sensible structure. A huge monorepo is only useful if you can point at the relevant part.
• Popularity is a hint, not proof. A 200-star repo with clean code beats a 20k-star repo of prompt collections.

Roles

• small, focused example apps and templates: code-truth
• awesome lists and tutorial repos with step-by-step folders: structure
• larger real-world projects: code-truth (how features combine). Name the interesting paths in reason.

Stars, the last push date and the pinned-version check belong in reason, for example: "12k stars, active, langgraph 1.x: multi-agent customer support with checkpointing in src/graph/."