📦 This repository is included in openclaw-skills — an aggregator repo with more skills. Star it to get all updates.
An OpenClaw Agent Skill that performs deep, multi-source analysis of any GitHub project and outputs a structured research report.
GitHub Explorer goes beyond a project's README. It automatically collects information from GitHub Issues, commits, Chinese developer communities (Zhihu, WeChat, V2EX), tech blogs (Medium, Dev.to), and Twitter/X discussions — then applies AI-driven judgment to produce a high-density, decision-ready report.
- Multi-source collection — Pulls from GitHub Issues, commits, Chinese communities (Zhihu / V2EX / WeChat), tech blogs (Medium / Dev.to), and Twitter discussions in parallel
- AI-driven judgment — Automatically classifies project stage (early experimental / rapid growth / mature stable / maintenance mode / stalled) and surfaces high-quality Issues
- Intent-aware search — Powered by search-layer v2; selects search strategy by context (project research, competitive comparison, latest updates, etc.) with results ranked by authority, freshness, and keyword relevance
- Competitive analysis — Identifies projects in the same space and produces a side-by-side comparison with links
- Knowledge graph check — Verifies coverage in DeepWiki, Zread.ai, and other knowledge sources
- Community sentiment — Quotes specific tweets and forum posts with source links; no vague summaries
- Structured output — Fixed report template for fast, consistent decision-making
npx skills add https://github.com/Arxchibobo/openclaw-github-explorercd ~/.openclaw/skills/
git clone https://github.com/Arxchibobo/openclaw-github-explorer github-explorerGitHub Explorer uses the following OpenClaw built-in tools (no extra install needed): web_search, web_fetch, browser
The following optional skills unlock stronger search and extraction capabilities. They are bundled in openclaw-search-skills:
| Skill | Purpose |
|---|---|
| search-layer | Multi-source search with intent-aware scoring (Brave + Exa + Tavily); v2 supports --intent, --queries, --freshness, --domain-boost |
| content-extract | High-fidelity content extraction; fallback for anti-scraping sites (WeChat, Zhihu, etc.) |
| mineru-extract | MinerU official API wrapper (downstream of content-extract) |
Install all dependencies at once:
# Tell your OpenClaw agent:
Install this skill: https://github.com/blessonism/openclaw-search-skillsOr manually:
git clone https://github.com/blessonism/openclaw-search-skills.git /tmp/openclaw-search-skills
cd ~/.openclaw/workspace/skills
ln -s /tmp/openclaw-search-skills/search-layer search-layer
ln -s /tmp/openclaw-search-skills/content-extract content-extract
ln -s /tmp/openclaw-search-skills/mineru-extract mineru-extractIf dependency skills are missing, github-explorer automatically falls back to built-in tools. Core functionality is not affected.
After installation, trigger the skill by asking your OpenClaw Agent in natural language:
Analyze the langchain project for me
Take a look at https://github.com/microsoft/graphrag
What's the deal with the ollama project?
The agent auto-triggers GitHub Explorer, runs multi-source collection and analysis, and outputs a complete research report.
The report title links to the GitHub repository. Sections:
| Section | Content |
|---|---|
| One-liner positioning | What the project is and what problem it solves |
| Core mechanism | Technical approach / architecture in plain language |
| Project health | Stars, Forks, License, team, commit trends, recent activity |
| Selected Issues | Top 3–5 high-quality Issues with links and key discussion points |
| Use cases | When you should use it |
| Limitations | When you should not use it |
| Competitive comparison | Similar projects with differences and links |
| Knowledge graph | DeepWiki / Zread.ai coverage status |
| Demo | Live demo links |
| Related papers | arXiv links |
| Community sentiment | Specific Twitter and Chinese community quotes with source links |
| Final judgment | Subjective assessment and recommendation |
- Full traceability — Every claim includes a source link so readers can verify
- Intent-aware search — Different collection phases use different search intents (
exploratoryfor research,comparisonfor competitive analysis,statusfor latest updates), with results weighted by authority, freshness, and relevance - Graceful degradation — When
web_fetchfails, automatically escalates tocontent-extract(MinerU) to maintain content quality - No vague claims — Community sentiment must cite specific content; summaries like "well received" are rejected
- Parallel collection — Multiple sources and queries run concurrently (
--queries) for efficiency
- OpenClaw agent runtime
- Optional: search-layer v2.1+, content-extract, mineru-extract (see openclaw-search-skills)
Made with by blessonism & Ms.Q