How fangbujituan/auto_test_platform works
ATP is a web-based automated testing platform. Users manage projects and test assets in the Vue client, or describe intent through AI; agent workflows can generate cases and scripts, invoke execution, and summarize results. API execution supports requests, variable handling, assertions, and reporting; UI execution uses Playwright. The platform also exposes capabilities through MCP and stores project, test, and result data. The README documents broader API/agent capabilities than the sampled implementation excerpts, so those links follow its stated architecture; the performance engine is planned, not an implemented workflow.
Main entry points: routes REST routes · engine.py API engine · models Domain models · business Business agents
Components 20
Web client
- Vue application
client/src/App.vue - Client API layer
client/src/api/request.js
Platform API
- REST routes
app/routes - Request schemas
app/schemas - AI service
app/services/ai_service.py - Execution services
app/services/executor.py
AI workflows
- Agent workflows
app/agents/workflows - Agent orchestrator
app/agents/orchestration/orchestrator.py - Business agents
app/agents/business - MCP server
app/mcp_server/server.py
Test execution
- API engine
app/engine/api_engine/engine.py - Sequence runner
app/engine/api_engine/runner/sequence_runner.py - HTTP client
app/engine/api_engine/http/client.py - Playwright UI engine
app/engine/ui_engine/playwright/executor.py - UI script manager
app/engine/ui_engine/playwright/script_manager.py
Persistence
- Domain models
app/models - Platform database
Other components
- User
- Tested services
- LLM provider
Connections 23
- User → Vue application: uses
- Vue application → Client API layer: requests
- Client API layer → REST routes: calls
- REST routes → Request schemas: validates
- REST routes → Execution services: invokes
- REST routes → AI service: invokes
- AI service → LLM provider: calls
- REST routes → Agent workflows: starts
- Agent workflows → Agent orchestrator: delegates
- Agent orchestrator → Business agents: dispatches
- Business agents → API engine: requests execution
- Business agents → Playwright UI engine: requests execution
- API engine → Sequence runner: runs collection
- Sequence runner → HTTP client: sends requests
- HTTP client → Tested services: exchanges HTTP
- Playwright UI engine → UI script manager: loads scripts
- Playwright UI engine → Tested services: automates browser
- Execution services → API engine: executes cases
- API engine → Domain models: reports results
- REST routes → Domain models: queries and updates
- Domain models → Platform database: persists
- MCP server → Execution services: invokes capabilities
- MCP server → Domain models: queries assets