Multi-Agent System

RepoDoctor AI

Paste any GitHub repo and 6 AI agents dissect it in real time. Architecture map, security audit, reliability report, code fixes, onboarding narrative, and a quiz.

View on GitHub Deploy your own Run locally
Heads up: this page is the project landing site. The actual app needs a server — GitHub Pages can host this page but cannot run the FastAPI backend. Deploy it to HF Spaces (free) or run it locally to use it.

The 6 agents

MAP

Mapper

Architecture, dependencies, entry points, language breakdown.

DOC

Storyteller

Plain-English narrative — how the codebase thinks.

SEC

Hacker

Security audit. SQLi, XSS, secrets, weak crypto, CWE-tagged.

SRE

SRE

Reliability. Bottlenecks, SPOFs, missing retries, scaling holes.

FIX

Fixer

Concrete code patches with before/after diffs.

QUIZ

Quizzer

5 onboarding MCQs that test real understanding.

One-click deploy

Hugging Face Spaces (free) Render Railway Fly.io
  1. Click a deploy button above. Sign in to that platform.
  2. Set the secret OPENAI_API_KEY in the platform env settings.
  3. Wait for build (~6 min first time). Single Docker image, no extra config.
  4. Open the URL the platform gives you. Done.

Run locally

git clone https://github.com/hammadshakeelai/RepoDoctor-Deploy
cd RepoDoctor-Deploy
copy .env.example .env
docker compose up --build
# open http://localhost:8000

Backend: FastAPI + OpenAI Agents SDK. Frontend: React + Vite. Single Docker container.

How a request flows

GitHub URL
    |
    v
FastAPI /api/analyze (Server-Sent Events)
    |
    +-> Cloner       (git clone --depth 1)
    +-> Mapper       (architecture, deps, entry points)
    +-> Storyteller  parallel
    +-> Hacker       parallel
    +-> SRE          parallel
    +-> Fixer        (code patches)
    +-> Quizzer      (onboarding MCQs)