github-trending

name	github-trending
description	Tracks trending GitHub repositories in crypto, DeFi, AI agents, and related domains
tags	["research","github","trending","developer","open-source"]
agent	researcher
var	${var} filters to a specific language or topic. If set (e.g., "rust", "AI agents"), narrow the scan. If empty, scan all relevant topics.

github-trending

Priority: P2 (runs daily) Schedule: 09:00 UTC Data sources: GitHub API, GitHub trending page Output: Trending repos report in memory/research/github/

Instructions

You are executing the github-trending skill for the Researcher agent.

Step 1: Load State

Read memory/research/github/trending-state.json:

{
  "last_scan": "ISO-8601",
  "tracked_repos": {
    "owner/repo": {
      "first_seen": "ISO-8601",
      "stars_when_found": 150,
      "stars_current": 890,
      "language": "Rust",
      "topic": "solana"
    }
  },
  "search_topics": ["solana", "defi", "mev", "blockchain", "smart-contract", "autonomous-agent", "llm-agent", "trading-bot", "zero-knowledge"]
}

Initialize with defaults if not found.

Step 2: Search Trending Repos

Run multiple GitHub searches for each topic:

curl -s -H "Authorization: token ${GITHUB_TOKEN}" \
  "https://api.github.com/search/repositories?q=topic:{topic}+created:>{30_days_ago}&sort=stars&order=desc&per_page=20"

Also search for recently active repos with high star velocity:

curl -s -H "Authorization: token ${GITHUB_TOKEN}" \
  "https://api.github.com/search/repositories?q=topic:{topic}+pushed:>{7_days_ago}+stars:>50&sort=stars&order=desc&per_page=20"

Step 3: Compute Star Velocity

For repos already in tracked_repos, calculate:

• Stars gained since last scan
• Star velocity (stars/day over last 7 days)
• Percentile rank within their category

Flag repos with:

• Breakout: >100 stars gained in 7 days AND star velocity accelerating
• Steady growth: >20 stars/day sustained over 2+ weeks
• New entrant: First seen this scan with >200 stars

Step 4: Analyze Top Repos

For top 10 repos by star velocity, fetch additional data:

curl -s -H "Authorization: token ${GITHUB_TOKEN}" "https://api.github.com/repos/{owner}/{repo}"
curl -s -H "Authorization: token ${GITHUB_TOKEN}" "https://api.github.com/repos/{owner}/{repo}/contributors?per_page=5"
curl -s -H "Authorization: token ${GITHUB_TOKEN}" "https://api.github.com/repos/{owner}/{repo}/commits?per_page=10"

Assess:

• Commit frequency and recency
• Number of contributors
• Quality of README and documentation
• License type
• Whether it is a fork of something notable
• Key dependencies (package.json, Cargo.toml, requirements.txt)

Step 5: Generate Report

{
  "scan_date": "2024-01-15",
  "trending_repos": [
    {
      "repo": "owner/repo",
      "url": "https://github.com/owner/repo",
      "description": "...",
      "language": "Rust",
      "stars": 890,
      "stars_7d_gain": 340,
      "star_velocity": 48.6,
      "forks": 120,
      "contributors": 15,
      "last_commit": "2024-01-14",
      "topics": ["solana", "defi"],
      "category": "breakout | steady_growth | new_entrant",
      "relevance": "Why this matters to our domain",
      "notable": true
    }
  ],
  "summary": {
    "total_repos_scanned": 200,
    "new_repos_found": 12,
    "breakout_repos": 3,
    "top_languages": {"Rust": 8, "TypeScript": 6, "Python": 4},
    "hottest_topics": ["AI agents", "restaking"]
  }
}

Step 6: Save and Notify

1. Write to memory/research/github/{YYYY-MM-DD}.json
2. Update memory/research/github/trending-state.json
3. Post breakout repos to analyst mesh
4. Notify on repos with >500 stars gained in 7 days

Anti-Patterns

• Do NOT count forks as trending. Only track original repositories.
• Do NOT include repos with no README or clear spam/scam projects.
• Do NOT exceed GitHub API rate limits. Space requests with 1s delay.
• Do NOT track private or archived repositories.

Exit Codes

• SKILL_OK — scan complete, N trending repos identified
• SKILL_PARTIAL — rate limit hit, partial results
• SKILL_EMPTY — no notable trending repos found
• SKILL_FAIL — could not access GitHub API

Output

Commit message format: researcher: github-trending — {N} repos, {M} breakouts [{top_repo}]