PromptShield MCP

Adversarial prompt detection & self-repair — FastAPI backend + React dashboard.

Proje Yapısı

promptshield_mcp/          ← backend (Python)
├── api.py
├── agentic_loop.py
├── config.py
├── requirements.txt
├── agents/
├── models/
├── utils/
├── mcp_tools/
├── dataset/data/
└── evaluation/

dashboard.jsx              ← React frontend (tek dosya)
render.yaml                ← Render deploy config
README.md

1 — Yerel Çalıştırma

Backend

cd promptshield_mcp
pip install -r requirements.txt
uvicorn api:app --reload

API: http://localhost:8000
Docs: http://localhost:8000/docs

Dashboard

npm create vite@latest promptshield-ui -- --template react
cd promptshield-ui
npm install
cp ../dashboard.jsx src/App.jsx
npm run dev

→ http://localhost:5173

2 — Render.com Ücretsiz Deploy

1. GitHub'a yükle

git init && git add . && git commit -m "initial"
git remote add origin https://github.com/KULLANICI/promptshield-mcp.git
git push -u origin main

2. render.com → New → Web Service → GitHub repoyu bağla

| Alan | Değer | |------|-------| | Root Directory | promptshield_mcp | | Runtime | Python 3 | | Build Command | pip install -r requirements.txt | | Start Command | uvicorn api:app --host 0.0.0.0 --port $PORT | | Plan | Free |

3. Deploy bittikten sonra dashboard.jsx içinde URL güncelle:

const API = "https://promptshield-mcp-xxxx.onrender.com";

API

| Method | URL | Ne yapar | |--------|-----|----------| | POST | /analyze | Sadece Detection Agent (hızlı) | | POST | /run | Tam 4-ajan pipeline | | GET | /dataset?split=test | Dataset kayıtları | | GET | /dataset/stats | İstatistikler | | GET | /evaluation/summary | Sistem karşılaştırması | | GET | /evaluation/categories | Kategori bazlı sonuçlar |

MCP Servers