
Healix is a monorepo containing two packages:
概览
Healix is a monorepo containing two packages: - — @healix/mcp (v2.0.0), an MCP server installed in developer IDEs. It orchestrates the full testing pipeline: auto-detect → explore → generate → execute → analyze → dashboard. - — A Next.js 16 / React 19 webapp deployed on Vercel. Hosts all AI calls (OpenAI server-side), the Supabase-backed dashboard, artifact storage, and the Inngest async generation pipeline. Users only need a HEALIX_API_KEY. All AI calls are proxied server-side through the webapp — no OpenAI key needed on user machines. Add to your MCP settings (~/.cursor/mcp.json, Claude Code, or Windsurf): Get your API key from the Healix dashboard → API Keys. 1. — local HTTP form collects test type, base URL, start command, PRD file (optional), and role credentials 2. — port, framework, and start command inferred from the project when not supplied 3.
README
以下为原文 README 节选,完整内容请查看原文。
Healix
End-to-end AI test generation and QA automation, from your IDE.
Healix is a monorepo containing two packages:
testbot-mcp/—@healix/mcp(v2.0.0), an MCP server installed in developer IDEs. It orchestrates the full testing pipeline: auto-detect → explore → generate → execute → analyze → dashboard.webapp/— A Next.js 16 / React 19 webapp deployed on Vercel. Hosts all AI calls (OpenAI server-side), the Supabase-backed dashboard, artifact storage, and the Inngest async generation pipeline.
Users only need a HEALIX_API_KEY. All AI calls are proxied server-side through the webapp — no OpenAI key needed on user machines.
Quick Start
1. Install the MCP server
npm install -g @healix/mcp
2. Configure your IDE
Add to your MCP settings (~/.cursor/mcp.json, Claude Code, or Windsurf):
{
"mcpServers": {
"healix": {
"command": "npx",
"args": ["@healix/mcp"],
"env": {
"HEALIX_API_KEY": "your-healix-api-key"
}
}
}
}
Get your API key from the Healix dashboard → API Keys.
3. Run a test
In your IDE prompt:
“Test my app using the healix mcp”
The pipeline runs automatically:
- Config UI — local HTTP form collects test type, base URL, start command, PRD file (optional), and role credentials
- Auto-detect — port, framework, and start command inferred from the project when not supplied
- Browser exploration — browser-use (Python subprocess) discovers real flows; falls back to a zero-dependency Playwright heuristic explorer if browser-use is unavailable
- PRD/AC parse — structured acceptance-criteria extraction via
/api/parse-prd; AC-less apps use exploration flows directly - Test generation — GPT-powered multi-agent fan-out via
/api/generate-tests; each test tagged[REQ:F#.S#.AC#] - Credential injection — per-role Playwright
storageStatewritten to.healix/auth-state-{role}.json - Tiered execution — three Playwright projects run in sequence:
- Tier A (
tierA-public) — unauthenticated / public flows - Tier B (
tierB-auth-{role}) — one project per role; markedblockedif login fails (A + C still run) - Tier C (
tierC-backend) — API/backend contract tests
- Tier A (
- Artifact upload — screenshots, videos, and Playwright traces uploaded to Supabase Storage
- Ingest — results POSTed to
/api/test-runs/ingest - Dashboard deep-link — opens the run page with tier pills, failure analysis, and artifacts
MCP Tools
| Tool | Description |
|---|---|
healix_test_my_app |
Full end-to-end pipeline. Accepts projectPath, baseURL, port, startCommand, testType (frontend/backend/both), prdFile, generateTests, openDashboard, credentials, codebaseContext, playwrightMcp, and more. |
healix_configure |
Opens the config UI and returns validated settings without running the pipeline. Useful for pre-flight checks. |
Configuration
安装
npx @healix/mcp配置
{
"mcpServers": {
"healix": {
"command": "npx",
"args": ["@healix/mcp"],
"env": {
"HEALIX_API_KEY": "your-healix-api-key"
}
}
}
}