Ask an AI agent to analyze open source? It clones the entire 500MB repo to
/tmp. Ask again? It clones again. Tokens wasted, time wasted. Use web search instead? Less accurate than reading the actual code, more hallucinations, and still burns tokens.Not anymore. Download only the folders you need, or extract raw links and use them directly.
🚀 Feed raw links into NotebookLM and build your own Deep Wiki from any GitHub repo.
| Input | → | Output | Use Case |
|---|---|---|---|
| GitHub specific folder | → | Raw URL list | Feed to NotebookLM, LLMs |
| GitHub specific folder | → | Downloaded files | Partial clone (skip the whole repo) |
curl -sL https://raw.githubusercontent.com/professional-ALFIE/github-extractor/master/install.sh | bashThis installs the skill and scripts to ~/.claude/skills/github-extractor/.
Get raw URLs for files in a specific GitHub path.
Why not git clone? You only need one folder, not 500MB of repo.
./extractor_githubRawLinks.py https://github.com/owner/repo/tree/main/docsOutput:
https://raw.githubusercontent.com/owner/repo/main/docs/guide.md
https://raw.githubusercontent.com/owner/repo/main/docs/api.md
- Auto-copied to clipboard (macOS)
- Supports any branch, tag, or commit
Requirements: Python 3 + requests
Tip: Set
GITHUB_TOKENto increase rate limit (60 → 5000 req/hour)
Download files from a specific GitHub directory — skip the full clone.
./extractor_githubRawFiles.py https://github.com/owner/repo/tree/main/docs ./local-docsAuto-ignores: node_modules/, .git/, __pycache__/, dist/, build/, .next/, venv/, .pyc, .exe, .dll, .so
Requirements: Python 3 + requests
Tip: Set
GITHUB_TOKENto increase rate limit (60 → 5000 req/hour)
Once installed, just ask:
- "Get raw links for https://github.com/owner/repo/tree/main/docs"
- "Download files from https://github.com/owner/repo/tree/main/src"
MIT