SuperAPI AI Directory
Partner with us中文AI newsFavorites
Skillsgithub-trending

github-trending

Data & analytics

View repository

Updated 10-15 12:29

About this skill

GitHub Trending Exploration

2026 Edition Core Understanding: GitHub Trending is already a list severely polluted by hype and AI slop (repeatedly confirmed by Karpathy and others).The only value of this skill is "structured data + strict filtering + writing a real angle", not outputting yet another list.

Scope of Responsibility(strictly followed):

  • Only responsible fordiscovery + trend analysis + daily report/insight report
  • Not responsible forprocessing the results into WeChat official account/Xiaohongshu/X posts (use /scout-to-article）
  • Not responsible formulti-AI deep competitor/community reaction research (use /multi-ai-research）

When to Trigger (Recommended Trigger Words)

  • Explore today's/this week's trending
  • "What good projects have there been recently in Rust / Python / Go"
  • "What is trending now in the AI / Agent / LLM field"
  • "Help me look at GitHub trending, are there any projects similar to X"
  • "What technologies have been growing rapidly recently and are worth paying attention to"
  • "Make a GitHub trending daily report / weekly report"
  • When you want to filter out hype and want real growth signals

Not suitable: pure quick fact queries (just let Claude answer directly).

Complete Workflow (must be in order, do not skip steps)

  1. Get clean data (prefer using scripts)

    # 在 skill 目录执行；如果 skill 已安装到 ~/.claude/skills，也可以换成对应安装路径
cd /path/to/spellbook/skills/github-trending

# 今日总榜 TOP 8
python3 scripts/fetch_trending.py --since daily --limit 8

# Rust 周榜
python3 scripts/fetch_trending.py --since weekly --language Rust --limit 6

# 开发者榜
python3 scripts/fetch_trending.py --developers --since daily --limit 10

    The script outputs structured JSON,never directly WebFetch raw HTML。

  2. Filter TOP N + filter out hype

    • Default 5-8
    • Prefer stars_today has real incremental growth + forks keep up + is not a pure awesome-list ones
    • Refer to reference/extended.md the "Hype Red Flags" in for quick exclusion

  3. Lightly enrich when necessary

    • Only for the final 2-3 projects that make the list, when necessary use WebFetch to look at the README header or Releases
    • Do not read the full README right away

  4. Apply template + must write Angle

    • Organize using the reference template
    • At the end there must be a separate paragraph "Today's Observation / Angle", this is where the output value lies

  5. Output + archive (optional)

    • Regular users: directly give a Markdown daily report
    • Want to post later: feed the JSON + angle structure to the downstream skill

Script Installation and Dependencies

Execute on first use:

python3 -m pip install -r requirements.txt

If there is no requirements.txt, install manually:

python3 -m pip install requests beautifulsoup4 lxml

Script location:scripts/fetch_trending.py(relative to this skill directory).

For supported parameters see the script --help. All errors will be returned as clear JSON, and will never fail silently.

Hard Output Requirements

The daily report must include an "Angle" paragraph at the end, otherwise it is considered unqualified output.

Characteristics of a good Angle:

  • Point out 1-2 cross-project patterns (technology evolution, community preferences, hype signals)
  • Give actionable recommendations ("this direction is worth following up, but prioritize looking at fork growth")
  • Based ontoday's real data, do not restate old knowledge

Advanced scenarios

  • Domain deep-dive report: first use a script to pull the monthly ranking by language, then combine it with reference/extended.md to write a comparison table using the evaluation framework.
  • Long-term tracking: use the /loop skill to run the script + analysis on a daily schedule, outputting to a fixed directory.
  • Want to know the community's real reaction: for key projects, also call /multi-ai-research。
  • Want to publish content: hand the structured results + angle to scout-to-article。

Extended Reference

All detailed frameworks (data sources, evaluation dimensions, hype identification, trend classification, good/bad Angle examples, scraper maintenance tips, 2026 community consensus) have been moved to reference/extended.md。

During daily triggers,no need toread it. Only load it when you want to customize templates, maintain scripts, or do high-precision analysis.

Update log: 2026-05 full refactor (introduced a reliable scraper layer + strict pipeline + split reference + removed all hardcoded dates/old trend lists). Data should always be based on the script's real-time output.