Features
Auto-Fix Agent
FixSense's AI agent automatically creates pull requests to fix failing E2E tests.
Overview
The Auto-Fix Agent is an AI-powered workflow that automatically generates code fixes for failing Playwright and Cypress tests. It uses AI to understand the failure context and create targeted fixes.
Setup
- Click "Enable Auto-Fix" in the FixSense dashboard
- FixSense creates a Pull Request adding the auto-fix workflow to your repo
- Review and merge the PR — this grants the workflow permission to run
- Done — auto-fix will trigger automatically on the next failure
You must merge the PR yourself — FixSense never pushes directly to your main branch.
How It Works
- Trigger — When FixSense detects a failure with high fix confidence, it triggers the auto-fix workflow
- Analysis — The AI reads the failure analysis, test code, and application code
- Fix generation — The agent modifies test code or test configuration to resolve the failure
- Verification — The agent runs your tests to confirm the fix works
- PR creation — A new branch (
fixsense/fix-*) is created with a Pull Request
BYOK (Bring Your Own Key)
Auto-fix uses your own AI API key. This means:
- Your code stays private — the agent runs in your CI runner
- You pay the AI provider directly — standard API rates, no FixSense markup
- Full control — set your own spending limits in your AI provider's console
Monthly Limits
| Plan | Auto-fixes/month |
|---|---|
| Free | 3 |
| Pro | 20 |
| Team | 100 |
Auto-fix limits exist to prevent runaway usage. If you consistently hit your limit, consider upgrading your plan.
What the Agent Can Fix
- Broken selectors — updated locators to match new DOM structure
- Timing issues — added proper waits and assertions
- API changes — updated test data or mock responses
- Assertion updates — corrected expected values after intentional changes
- Test config issues — headless mode, timeouts, retries, baseURL in
playwright.config.ts
What the Agent Won't Do
- Modify application source code
- Push directly to your main branch (always creates a separate branch + PR)
- Run without your explicit API key configuration
- Access external services or databases
Security
The auto-fix agent operates within strict boundaries:
- Runs inside your GitHub Actions runner
- Only creates branches prefixed with
fixsense/fix-* - Can modify test files and
playwright.config.ts— never application code, secrets, or CI workflows - All changes are reviewed via standard PR process