RAG Chat Over Git History
Ask your repo questions in plain English. Scrapecat retrieves the most relevant commits and summarizes them at a feature level.
What it does
Stop digging through git logs. Ask questions and get structured answers.
RAG Chat
Ask questions about your code history in natural language. Vector search finds the most relevant commits and summarizes them with an LLM.
Feature-Level Summaries
Related commits are grouped by feature, bug fix, or refactor. No more reading individual commit messages.
Source Citations
Every AI response includes collapsible source links to the exact GitHub commits used to generate the answer.
Report Artifacts
Ask for a report and get a styled card with structured output. Perfect for standups and stakeholder updates.
Importance Reranking
Commits are scored by type, file count, and PR status. The 30 most important ones are surfaced first.
Run Locally
Works with OpenRouter or Ollama. Your code never leaves your machine if you use the local LLM option.
FAQ
No. It treats your data as read-only as it only reads metadata and commit history to build reports.
OpenRouter (Gemma 4, DeepSeek, GPT-4o, etc.) or Ollama for fully local inference. No API key needed for Ollama.
Not yet. Scrapecat is self-hosted via Docker or local setup. Your data stays on your machine.