MCP Presentations

# Testing Guide ## Pre-Deployment Testing ### 1. Type Check ```powershell npm run type-check ``` ✅ Expected: No errors ### 2. Lint Check ```powershell npm run lint:fix ``` ✅ Expected: Code formatted properly ### 3. Local Development Test ```powershell # Terminal 1 cd python-worker wrangler dev --port 8788 # Terminal 2 wrangler dev --port 8787 ``` ✅ Expected: Both workers start successfully ## Deployment Testing ### 1. Deploy Python Worker ```powershell cd python-worker wrangler deploy ``` ✅ Expected output: ``` Published mcp-presentations-python (X.XX sec) https://mcp-presentations-python.<account>.workers.dev ``` ### 2. Deploy Main Worker ```powershell cd .. wrangler deploy ``` ✅ Expected output: ``` Published mcp-presentations (X.XX sec) https://mcp-presentations.<account>.workers.dev ``` ## Functional Testing ### Test 1: Basic Connection **Tool**: Browser or curl ```powershell curl https://mcp-presentations.<account>.workers.dev/sse ``` ✅ Expected: SSE connection established (streaming response) ### Test 2: Calculator Tools (via AI Playground) 1. Go to: https://playground.ai.cloudflare.com/ 2. Enter: `https://mcp-presentations.<account>.workers.dev/sse` 3. Click Connect ✅ Expected: See 3 tools available: - add - calculate - create_presentation **Test add:** - Ask: "What's 5 + 3?" - ✅ Expected: Returns 8 **Test calculate:** - Ask: "Calculate 6 times 7" - ✅ Expected: Returns 42 ### Test 3: PowerPoint Creation - Simple **Prompt**: "Create a 1-slide presentation with title 'Hello World'" ✅ Expected response: ``` ✅ PowerPoint presentation "Hello World.pptx" created successfully with 1 slide(s)! File size: XX.XX KB The presentation has been generated. ``` ### Test 4: PowerPoint Creation - Complex **Prompt**: "Create a 3-slide presentation about cats with bullet points" ✅ Expected: - Slide 1: Title slide about cats - Slide 2-3: Content slides with bullet points - Success message with file size ### Test 5: Error Handling - Empty Slides Send request with no slides: ```json { "title": "Empty", "slides": [] } ``` ✅ Expected: Error message about empty slides ### Test 6: Error Handling - Invalid Layout ```json { "title": "Test", "slides": [ {"layout": "invalid_layout", "title": "Test"} ] } ``` ✅ Expected: Either validation error or defaults to title_and_content ## Integration Testing ### Test 7: Claude Desktop Integration 1. Edit Claude config: ```json { "mcpServers": { "presentations": { "command": "npx", "args": ["mcp-remote", "https://mcp-presentations.<account>.workers.dev/sse"] } } } ``` 2. Restart Claude Desktop 3. Check tools: Type `@` and look for presentations server ✅ Expected: - presentations server shows up - 3 tools visible (add, calculate, create_presentation) 4. Test: "Create a presentation about dogs" ✅ Expected: AI creates and confirms presentation ## Performance Testing ### Test 8: Concurrent Requests Use a load testing tool or script: ```powershell # Simple PowerShell concurrent test 1..10 | ForEach-Object -Parallel { # Make request to create presentation Write-Host "Request $_" } ``` ✅ Expected: All requests succeed ### Test 9: Large Presentation **Prompt**: "Create a 20-slide presentation about world history" ✅ Expected: - Generates successfully - File size proportional to slides - Reasonable response time (<30 seconds) ### Test 10: File Size Limits Create presentation with many slides and long content. ✅ Expected: - Works up to reasonable sizes (~5MB) - Fails gracefully if too large ## Service Binding Testing ### Test 11: Python Worker Direct Call (Should Fail) ```powershell curl https://mcp-presentations-python.<account>.workers.dev/ ``` ✅ Expected: Worker responds (it's actually public by default) ⚠️ Note: In production, you might want to add authentication ### Test 12: Service Binding Verification Check logs while making a request: ```powershell # Terminal 1: Watch main worker wrangler tail mcp-presentations # Terminal 2: Watch python worker wrangler tail mcp-presentations-python # Terminal 3: Make a request # (use AI Playground) ``` ✅ Expected: See logs in both workers showing the request flow ## Monitoring Tests ### Test 13: Log Inspection ```powershell # Main worker logs wrangler tail mcp-presentations # Python worker logs wrangler tail mcp-presentations-python ``` ✅ Expected: See request logs, no errors ### Test 14: Deployment Status ```powershell wrangler deployments list ``` ✅ Expected: Both workers show latest deployments ## Edge Cases ### Test 15: Special Characters Create presentation with title: `Test's "Presentation" & More` ✅ Expected: Handles special characters properly ### Test 16: Unicode Content Create presentation with emoji and international characters: ``` Title: "🎉 International Présentation 日本語" Bullets: ["Emoji 😀", "Français", "日本語"] ``` ✅ Expected: Renders correctly in PowerPoint ### Test 17: Very Long Content Create slide with 100+ character bullet points. ✅ Expected: Handles gracefully (might truncate or wrap) ### Test 18: Empty Title ```json { "title": "", "slides": [{"layout": "title", "title": "Test"}] } ``` ✅ Expected: Uses default name or empty string ## Regression Testing ### Test 19: Existing Tools Still Work After deployment, verify: - add tool still works - calculate tool still works ✅ Expected: No impact on existing functionality ### Test 20: Multiple Sequential Requests Create 5 presentations in a row. ✅ Expected: All succeed, no state leakage ## Security Testing ### Test 21: Authentication (Currently None) Try accessing without any credentials. ✅ Expected: Works (this is an authless server) ⚠️ Note: For production, consider adding auth ### Test 22: Injection Attempts Try malicious input in titles/bullets: ```json { "title": "<script>alert('xss')</script>", "slides": [...] } ``` ✅ Expected: Safely handled (no code execution) ### Test 23: Resource Exhaustion Try creating 1000-slide presentation. ✅ Expected: Either succeeds or fails gracefully with error ## Documentation Testing ### Test 24: Follow Deployment Guide Have someone unfamiliar follow DEPLOYMENT.md. ✅ Expected: They can successfully deploy both workers ### Test 25: Example Verification Test all examples from EXAMPLES.md. ✅ Expected: All examples work as documented ## Cleanup Testing ### Test 26: Undeploy and Redeploy ```powershell # Delete deployments wrangler delete mcp-presentations-python wrangler delete mcp-presentations # Redeploy .\deploy.ps1 ``` ✅ Expected: Clean redeploy works ## Final Checklist Before declaring "Done": - [ ] Type check passes - [ ] Both workers deploy successfully - [ ] All 3 tools work in AI Playground - [ ] Can create simple presentation - [ ] Can create complex presentation - [ ] Error handling works - [ ] Claude Desktop connection works - [ ] Logs show proper flow - [ ] Documentation is accurate - [ ] No security vulnerabilities - [ ] Performance is acceptable ## Automated Test Script (Future Enhancement) ```powershell # test-all.ps1 (not yet implemented) # This could automate all the above tests Write-Host "Running full test suite..." # Type check npm run type-check # Deploy to test environment # Run functional tests # Run integration tests # Run performance tests # Generate report Write-Host "All tests passed! ✅" ``` ## Known Issues / Limitations 1. **No file download**: Currently returns success message only, no direct file download 2. **No authentication**: Anyone with URL can use the service 3. **No rate limiting**: Beyond Cloudflare's automatic limits 4. **Limited layouts**: Only 3 layout types supported 5. **No images**: Cannot add images to slides yet 6. **No themes**: Uses default PowerPoint theme 7. **No custom fonts**: Uses default fonts ## Future Testing Needs When adding new features, test: - [ ] Image uploads - [ ] Custom themes - [ ] More layouts - [ ] Templates - [ ] Authentication - [ ] Rate limiting - [ ] File storage (R2/KV) - [ ] Analytics --- **Testing Status**: ✅ Ready for production (with noted limitations)