weapp-pilot-mcp

English | 简体中文

WeApp Pilot MCP is a local MCP server for LLM and Agent workflows that automate WeChat Mini Programs through WeChat DevTools.

The current public scope focuses on route-aware smoke validation, structured scenario execution, and runtime diagnostics for Taro-based projects.

• Taro-first project detection and build orchestration
• WeChat DevTools plus miniprogram-automator runtime control
• Read-only readiness checks with env_check
• Active startup validation with preflight_check
• Route-aware baseline validation with project_routes, route_search, and smoke_run
• Structured step execution with scenario_run
• Default-safe redaction for persisted console and network artifacts

• local-machine execution
• WeChat DevTools must already be installed and logged in
• one active runtime session per MCP process
• Native Mini Program and UniApp adapters are not implemented yet

• @weapp-pilot/shared: contracts, error model, and tool schemas
• @weapp-pilot/weapp-core: local orchestration, runtime control, assertions, and artifacts
• @weapp-pilot/mcp-server: stdio MCP server and tool registry

pnpm install
pnpm build
pnpm test
pnpm smoke:fixture

pnpm smoke:fixture runs the fake end-to-end validation flow around the bundled Taro fixture and proves that the MCP contracts and runtime orchestration are wired correctly before touching a real WeChat DevTools session.

Use a direct Node entrypoint when wiring the workspace build into an MCP client so stdio is not polluted by package-manager script banners:

{
  "mcpServers": {
    "weapp-pilot": {
      "command": "node",
      "args": ["<repo-root>/packages/mcp-server/dist/bin.js"]
    }
  }
}

The prepared public package name is weapp-pilot-mcp:

npx -y weapp-pilot-mcp

Typical published-package MCP wiring: