AI Site Detection

中文文档

Detect whether an API gateway supports OpenAI, Anthropic, or Gemini-compatible endpoints — by probing real HTTP requests.

Screenshots

Home — Enter Base URL and API Key:

Detection Results — Provider Cards with Re-detect:

Request Trace Drawer — Masked HTTP Details:

Features

• Multi-provider detection — OpenAI Chat, OpenAI Codex (Responses API), Anthropic, and Gemini
• Single provider re-detection — Re-detect any individual provider without re-running the full scan
• Confidence scoring — High / Medium / Low based on model list + endpoint probe results
• Request tracing — Full masked HTTP traces for every probe (headers, body, response)
• Smart Paste — Paste text containing a URL and API key, auto-extract both fields
• Two deployment modes — Full-stack (Vue + Fastify) or standalone Cloudflare Worker

Architecture

├── shared/          # Shared TypeScript types (DetectRequest, ProviderDetectionResult, etc.)
├── server/          # Fastify API server (TypeScript)
│   └── src/
│       ├── providers/   # OpenAI / Anthropic / Gemini detector implementations
│       ├── services/    # detectAll / detectOne orchestrators
│       └── routes/      # POST /api/detect, POST /api/detect-one
├── web/             # Vue 3 + Vite frontend
│   └── src/
│       ├── api/         # API client (detectSite, detectOne)
│       └── components/  # DetectForm, ProviderCard, ResultSummary, RequestTraceDrawer
└── worker/          # Cloudflare Worker (single-file, self-contained)
    └── worker.js

Quick Start

Full-Stack Mode (Vue + Fastify)

# Install dependencies
npm install

# Start both frontend and backend in dev mode
make dev

# Or start them separately
make web     # Frontend only (Vite dev server on port 5173)
make server  # Backend only (Fastify on port 3000)

Cloudflare Worker Mode

cd worker
npm install
npx wrangler dev worker.js

Docker Deployment

# Build and start
cd deploy
docker-compose up -d

# View logs
docker-compose logs -f

# Stop
docker-compose down

Access at: http://localhost:3000

Custom port: Edit deploy/docker-compose.yml:

ports:
  - "8080:3000"  # Map host port 8080 to container port 3000

Environment variables:

Variable	Default	Description
PORT	3000	Server port
HOST	0.0.0.0	Server host
NODE_ENV	production	Node environment

API

POST /api/detect

Full detection across all providers.

Request:

{
  "baseUrl": "https://api.example.com",
  "apiKey": "sk-...",
  "timeoutMs": 8000
}

Response:

{
  "ok": true,
  "normalizedBaseUrl": "https://api.example.com",
  "startedAt": "...",
  "finishedAt": "...",
  "results": [
    {
      "provider": "openai-chat",
      "supported": true,
      "confidence": "high",
      "models": ["gpt-4o-mini", "..."],
      "endpointTried": "https://api.example.com/v1/chat/completions",
      "statusCode": 200,
      "latencyMs": 342,
      "traces": [...]
    }
  ]
}

POST /api/detect-one

Re-detect a single provider.

Request:

{
  "baseUrl": "https://api.example.com",
  "apiKey": "sk-...",
  "provider": "anthropic",
  "timeoutMs": 8000
}

Response: A single ProviderDetectionResult object (same shape as one element in the results array above).

Scripts

Command	Description
make dev	Start frontend + backend concurrently
make web	Start frontend dev server
make server	Start backend dev server
make build	Build both frontend and backend
make test	Run all tests

Tech Stack

• Frontend: Vue 3 + TypeScript + Vite
• Backend: Fastify + TypeScript
• Worker: Cloudflare Workers (vanilla JS)
• Testing: Vitest

License

MIT