⚒️ Transform GitHub repos into evidence-based architecture blueprints with Claude

54 views · ⚒️ Engineering

💡 Pro Tip — If you need GitHub data beyond what the REST API gives you, ScraperNode has a repository scraper that extracts metadata at scale without token rate limits.

View All Scrapers

Description

Requirements: GitHub API token (repo scope), Anthropic API key (Claude Sonnet 4.5), Slack Bot Token (optional)

This workflow analyzes any public GitHub repository and generates an evidence-based architecture blueprint — with Mermaid.js diagrams, technical stack analysis, and risk assessment — then pushes it as README_ARCH.md directly to the repo.

What makes it different: A strict anti-hallucination prompt ensures Claude only describes technologies that actually exist in the code. No Dockerfile? Docker is never mentioned. No .tf files? Terraform doesn’t appear. Every claim traces back to real file evidence.

How it works

  1. A GitHub URL is submitted via the web form or webhook API
  2. The workflow fetches repo metadata, the full file tree, and contents of key files (package.json, requirements.txt, Dockerfiles, entry points, etc.)
  3. Claude Sonnet 4.5 (temperature 0.1) analyzes only the evidence — strict rules prevent inventing technologies not found in the code
  4. A Markdown blueprint with architecture diagrams and risk analysis is assembled, auto-fixed, and pushed to the repo’s default branch
  5. A Slack notification is sent and the form user receives a styled success page

Example output

The generated README_ARCH.md includes:

Use cases

Setup

See the Setup & Overview sticky note in the workflow for step-by-step credential configuration.

🔗 Nodes Used

HTTP Request, Slack, Webhook, Basic LLM Chain, Anthropic Chat Model, n8n Form Trigger

📥 Import

Download workflow.json and import into n8n: Workflow menu → Import from File

📖 Importing guide · 🔑 Credential setup