Kairo

Kairo is an MCP server that closes the loop between code generation and test verification.

It autonomously analyzes a project → generates Playwright tests via OpenAI → runs them → classifies failures → recommends fixes — iterating up to 3 times.

How it works

analyze → generate tests → run → classify → fix → rerun (max 3x)

Exit conditions: all pass · iteration 3 reached · only flaky/env failures · app unresponsive

Install

npm install -g @kairo/mcp

Or use via npx (no install needed):

{
  "mcpServers": {
    "kairo": {
      "command": "npx",
      "args": ["-y", "@kairo/mcp"],
      "env": {
        "OPENAI_API_KEY": "sk-..."
      }
    }
  }
}

MCP Tools

Tool	Description
analyze_project	Detect stack, routes, endpoints, auth strategy
generate_tests	Generate Playwright tests via OpenAI
run_tests	Execute with double-run flakiness detection
classify_failures	bug / flaky / env / security
get_fixes	Fix recommendations with confidence levels

Dashboard

Start the API:

cd apps/api && pnpm dev

Start the dashboard:

cd apps/dashboard && pnpm dev
# → http://localhost:5173

Development

pnpm install
pnpm typecheck     # typecheck all packages
pnpm build         # build all packages
pnpm test          # run all tests

Environment Variables

Variable	Required	Description
OPENAI_API_KEY	Yes	OpenAI API key for test generation and fix recommendations
DATABASE_URL	No	SQLite path (default: ./kairo.db)
PORT	No	API port (default: 3000)

Stack

• packages/analyzer — Project analysis (stack, routes, auth)
• packages/generator — OpenAI-powered test generation
• packages/runner — Playwright executor with double-run
• packages/classifier — Deterministic failure classification
• packages/fixer — OpenAI fix recommendations
• packages/mcp — MCP server + orchestration loop
• apps/api — Express REST API + SQLite
• apps/dashboard — React + Vite + HeroUI dashboard