| name | github-trends-ops |
| description | Fetch and analyze GitHub trending repositories (daily/weekly) to surface high-signal development trends. |
| memory_tags | ["domain:development","workflow:trend-analysis","skill_role:researcher","repo_boundary:tools","source:github-trending","risk:low"] |
GitHub Trending Operations
Use this skill to monitor what the open-source community is building in real-time.
Trigger
Use when:
- the user asks "what's trending on GitHub today?" or "what's big this week?"
- starting the morning debrief to ground daily planning in global developer signals
- performing market research for new product ideas
Workflow
- Define the candidate universe:
- Record the window (
daily or weekly), language filter, fetch time, and GitHub surfaces included.
- For a normal Trending scan, treat the complete fetched Trending page as the minimum candidate universe; do not begin from remembered repositories or a personalized feed.
- If the user asks what is broadly breaking out beyond Trending, add explicit GitHub Search/topic/release surfaces and label that expanded scope.
- Fetch Trends:
- Daily:
python3 georgeskills/skills/github-trends-ops/scripts/fetch_github_trends.py --since daily
- Weekly:
python3 georgeskills/skills/github-trends-ops/scripts/fetch_github_trends.py --since weekly
- Resolve primary sources:
- Open each shortlisted repository and use its canonical repository, README, releases, and linked project documentation for identity and claims.
- Cluster renamed, forked, or duplicate implementations instead of counting them as separate trends.
- Treat posts and newsletters as discovery context, not as substitutes for the canonical repository.
- Analyze:
- Look for clusters of technology (e.g., "many new AI agents", "rise in Rust-based infra").
- Check whether top repositories align with the user's stated projects or research lanes.
- Label unavailable language pages, rate limits, failed fetches, or missing historical counters as source gaps rather than zero signal.
- Run the coverage gate before ranking:
- Confirm every repository above the chosen rank/star-growth threshold was either shortlisted or explicitly omitted.
- Record a compact omitted-candidate audit with repository, observed signal, and reason (
duplicate, off-scope, low evidence, or already covered).
- Log:
- Summary is logged to
<private-repo>/captures/social-media/github/trends/YYYY-MM-DD.md.
Output Contract
Return the snapshot time and scope, ranked repositories/clusters with canonical
links, the most important source gaps, and the omitted-candidate audit. A scan
is not complete when a high-signal candidate silently disappears between fetch
and ranking.