🐙 GitHub to Context — Repo to Markdown for LLMs

Turn any GitHub repository into a single, AI-ready Markdown context file in seconds.
Perfect for ChatGPT / Claude / RAG pipelines / codebase-aware agents.

Most AI workflows fail not because the model is weak, but because the context is incomplete. Copy-pasting files is painful, cloning is slow, and you still need to manually prune binaries, images, and lock files. GitHub to Context solves this by using the GitHub API to fetch the repository tree, filter unwanted files, and concatenate the codebase into a clean, structured Markdown output ready for ingestion or pasting into an LLM.

---

✅ What this Actor delivers

• Single Markdown context (GFM-friendly) containing:
  • Repository metadata (repo + branch)
  • File structure overview
  • Full file contents, separated and labeled
• Auto-filtering for non-LLM-friendly content (binaries, media, lock files)
• Private repo support via GitHub Token (recommended fine-grained token)
• Token-awareness: estimates token usage so you can control cost and context size
• RAG-ready dataset: each run produces structured output suitable for vector DB ingestion

---

✨ Why use this Actor?

Feature	❌ Local Cloning	✅ GitHub to Context Actor
Speed	Slow (downloads history & assets)	Instant (API Tree Fetching)
Output	Files & folders	Single Markdown file
Filtering	Manual .gitignore work	Auto-ignores locks, images, binaries
Context	Hard to paste into an LLM	Ready to paste / embed
Access	Public repos only	Public & Private (via token)

---

🧠 Typical workflows

1) Paste into ChatGPT / Claude

Use the output directly as a “codebase context” prompt, then ask:

• “Find security issues”
• “Explain architecture”
• “Generate a refactor plan”
• “Add tests for X module”

2) RAG pipeline ingestion

Store Markdown output into Pinecone / Qdrant / Weaviate / OpenAI Vector Store:

• Chunk by file boundaries
• Add metadata: repo, branch, path
• Retrieve relevant slices for agent reasoning

3) Documentation generation

Have an LLM generate:

• README / Wiki pages
• API docs / onboarding guides
• ADRs (Architecture Decision Records)

---

📦 Output format

The Actor produces a dataset item containing:

• repo — repository identifier (ex: apify/crawlee)
• branch — scraped branch
• fileCount — how many files were included
• totalTokens — estimated token size of the generated context
• scrapedAt — ISO timestamp
• markdown (or similar field in your implementation) — the full concatenated Markdown context

Example Markdown output structure

# Repository: apify/crawlee
# Branch: master

## File Structure
src/index.ts
src/utils.ts
...

## File: src/index.ts
```typescript
export * from './crawlee';
...