Turn GitHub Traffic into an evidence-backed growth story.

GitHub Repo Insights is an agent skill for understanding how a repository is actually growing. It combines GitHub Traffic, star events, referrers, popular paths, releases, and community signals—then keeps observed facts separate from interpretation.

What it answers

Question	Evidence used
Are more people discovering the project?	14-day views and unique visitors
Are they interested enough to try it?	Clones and unique cloners
Is growth accelerating or fading?	Recent 7 days vs. previous 7 days, daily peaks, star events
Where does attention come from?	GitHub, search, social, community, and other referrers
What do visitors care about?	README, docs, releases, issues, and other popular paths
What should we improve next?	Evidence-linked onboarding, localization, release, and distribution actions

Install

npx skills@latest add awesome-skills/github-repo-insights -g -y

Requirements: Python 3.10+ and an authenticated GitHub CLI.

gh auth status

Repository Traffic requires push access. For public repositories without push access, the skill still reports public metadata, star momentum, releases, and community signals—and explicitly marks private Traffic data as unavailable instead of pretending it is zero.

Ask your agent

Use $github-repo-insights to analyze why this repository is growing.

Compare the latest 7 days with the previous 7 days and tell me what to improve next.

Where did this repository's traffic come from, and which pages are people reading?

Or collect a normalized JSON snapshot directly:

python3 scripts/collect_github_repo_insights.py OWNER/REPO \
  --output /tmp/github-repo-insights.json

Inside a local Git checkout, OWNER/REPO can be omitted and detected automatically.

Example snapshot

An analysis of tt-a1i/archify for 2026-07-02 → 2026-07-15 found:

Signal	Result
Views	23,877
Unique visitors	11,062
Clones	6,902
Unique cloners	4,320
Peak daily stars	552 on July 15

The strongest recorded referrers were GitHub, X / t.co, and Google. The Chinese README was already the repository's third most-viewed path, showing that localization was contributing measurable demand.

This is a dated example, not a live counter. GitHub Traffic exposes a rolling 14-day window, so recurring monitoring should save snapshots over time.

Built for honest reporting

• Window totals stay totals. Daily unique visitors and unique cloners are never summed.
• Unavailable is not zero. Missing permissions and incomplete history remain visible.
• Ratios stay directional. Stars-per-view and clones-per-view are not presented as user conversion rates.
• Correlation is not causation. Releases, posts, Trending appearances, and traffic spikes are connected only when dates and evidence support the inference.
• Next steps follow the bottleneck. Recommendations are tied to acquisition, onboarding, localization, packaging, or retention signals found in the data.

What gets collected

GitHub REST API      repository · traffic · referrers · popular paths · releases
GitHub GraphQL API   timestamped star events and daily momentum
Collector            normalized JSON · 7d comparisons · peak dates · warnings
Agent skill           evidence-backed interpretation and next actions

The metric definitions and important GitHub API limitations are documented in references/metrics.md.

License

MIT © Awesome Skills