AI Skill Report Card
Analyzing Sentry Logs for Playwright
Quick Start15 / 15
REPRODUCTION BLUEPRINT
Environment: https://app.example.com/dashboard | Chrome 120 | 1920x1080
Initial State: Authenticated user (ID: user_123), dark_mode=true
Execution Steps:
1. Click "Export Data" button
2. Select "CSV Format" from dropdown
3. Submit form triggers POST /api/export
Assertion Point: ReferenceError: exportManager is not defined
Mocking Requirements: Mock POST /api/export to return 500 Internal Server Error
Recommendation▾
Compress the workflow section - the progress checklist and detailed steps are somewhat redundant
Workflow15 / 15
Progress:
- Extract environment context from Tags
- Identify initial application state from Context
- Map breadcrumbs to chronological actions
- Locate the exact error point
- Flag required API mocks from failed requests
- Format as structured blueprint
Step-by-Step Process
-
Parse Environment Data
- Extract URL, browser, OS from Tags section
- Note viewport size if available
- Capture release version
-
Analyze Application State
- User authentication status from Context
- Feature flags or configuration from Tags
- LocalStorage/Cookie dependencies
-
Convert Breadcrumbs to Actions
- Focus on:
ui.click,ui.input,navigation,http - Map UI events to user actions
- Note failed HTTP requests (4xx/5xx status)
- Focus on:
-
Identify Assertion Point
- Extract exact exception name and message
- Note line numbers from stack trace
- Determine observable failure state
-
Document Mock Requirements
- List failed API endpoints
- Include expected error responses
- Note timing dependencies
Recommendation▾
Add an edge case example showing how to handle logs with no breadcrumbs or minimal context
Examples18 / 20
Example 1: Input: Sentry log with breadcrumbs showing navigation to /checkout, form submission, and TypeError on payment processing Output:
REPRODUCTION BLUEPRINT
Environment: https://shop.com/checkout | Firefox 118 | Mobile 375x667
Initial State: Cart contains 2 items, guest checkout
Execution Steps:
1. Navigate to checkout page
2. Fill shipping form fields
3. Click "Continue to Payment"
4. Enter credit card details
5. Submit payment form
Assertion Point: TypeError: Cannot read property 'validate' of null
Mocking Requirements: Mock POST /api/payment/validate to return 422 validation error
Example 2: Input: Insufficient breadcrumb data, only shows final error without user actions Output:
REPRODUCTION BLUEPRINT
Environment: https://admin.app.com/users | Chrome 119 | 1440x900
Initial State: Admin user authenticated
Execution Steps:
INSUFFICIENT DATA FOR: User interaction sequence leading to error
Assertion Point: ReferenceError: userManager is not defined at line 247
Mocking Requirements: None identified
Recommendation▾
Include a template section with a blank blueprint format that users can copy and fill
Best Practices
- Extract only explicit data - Never invent selectors or API endpoints not shown in logs
- Preserve exact error messages - Copy exception text verbatim for assertion matching
- Flag data gaps - Use "INSUFFICIENT DATA FOR [X]" when logs incomplete
- Map HTTP failures - Any 4xx/5xx response needs mocking strategy
- Include timing context - Note if errors occur after delays or async operations
- Document prerequisites - Authentication, feature flags, or setup requirements
Common Pitfalls
- Don't guess CSS selectors - Use descriptive text like "Login button" instead of "#login-btn"
- Don't assume user paths - Multiple scenarios may lead to same error
- Don't ignore environment differences - Mobile vs desktop can affect reproduction
- Don't skip failed requests - Missing mocks cause different test failures
- Don't paraphrase errors - Exact exception text needed for reliable assertions
- Don't merge unrelated events - Separate distinct user workflows into scenarios