AI Fixes

Every competitor stops at "here's what changed." Frontguard goes further: here's a fix, and it re-rendered the page with the fix applied to confirm it works. Over time it learns which fixes you accept and reuses them.

Verifying fixes in a sandbox

A suggested fix is only useful if it works. With verifyFixes: true, Frontguard applies the patch, re-renders, and re-compares:

1Apply the generated CSS patch in a sandbox.

2Re-render the page with the patch injected.

3Re-compare the result against the baseline.

4Mark ✅ Verified if within threshold, ⚠️ Unverified otherwise.

frontguard.config.ts

generateFixes: true,
verifyFixes: true,
fixSandbox: 'local', // 'local' | 'daytona'

The fix-pattern database

Frontguard keeps a local SQLite store of the fixes you accept and reject. The more you accept, the more often it reuses a known-good pattern instead of asking the AI again.

$ frontguard accept-fix <id> # positive training signal
$ frontguard reject-fix <id> # negative signal
$ frontguard export-patterns > fix-patterns.json

Learning loop

A pattern is reused once it has been accepted ≥3 times with no rejections — so a one-off accept never overrides the model. Verified fixes are recorded as accepted automatically.