# MCP Server Testing Automation
This directory contains automated tests that simulate MCP Inspector usage.
## Quick Start
### Run Automated HTTP Tests
Test the running MCP HTTP server:
```bash
# Server must be running on http://localhost:8000
uv run python scripts/test_mcp_http.py
# Or test a different URL
uv run python scripts/test_mcp_http.py --url http://localhost:8000
```
### What Gets Tested
The automated tests cover:
1. **Health Endpoint** - Verifies server is responsive
2. **Metrics Endpoint** - Checks monitoring data
3. **SSE Endpoint** - Confirms SSE endpoint is reachable
4. **Messages Endpoint** - Validates request handling
## Integration with CI/CD
The tests run automatically on every push via GitHub Actions (`.github/workflows/test.yml`):
```yaml
- Unit tests with pytest
- Start HTTP server
- Run integration tests
- Report results
```
## Manual Inspector Testing
For full interactive testing with MCP Inspector:
```bash
# Terminal 1: Start server
docker-compose up
# Terminal 2: Start MCP Inspector
npx @modelcontextprotocol/inspector
# Configure Inspector:
# - URL: http://localhost:8000/sse
# - Transport: SSE
```
## Extending the Tests
To add more automated tests, edit `scripts/test_mcp_http.py`:
```python
async def test_my_feature(self) -> bool:
"""Test description."""
print("🧪 Testing my feature...")
# Your test code here
return True
```
Then add it to `run_all_tests()`:
```python
results['my_feature'] = await self.test_my_feature()
```
## Limitations
The automated tests currently focus on:
- HTTP endpoint availability
- Basic request/response validation
- Server health checks
For full MCP protocol testing (tools, resources, prompts), use:
- Existing pytest integration tests in `tests/`
- Manual testing with MCP Inspector for UX validation
- The MCP Inspector provides the best interactive testing experience
## Future Enhancements
Potential additions:
- Full SSE connection simulation
- Automated tool invocation tests
- Resource access validation
- Performance benchmarking
- Load testing scenarios
