View repo HH

Self-healing Playwright operations

Helix Heal fixes broken Playwright locators before they stall the release.

Helix Heal reads failed reports, trace evidence, and DOM snapshots, then returns confidence-ranked locator repairs as reviewable dry-run patches. Local-first by default. No account. No artifact upload.

Run the free CLI Watch fail to green
npx helix-heal analyze --report playwright-report.json --trace test-results Works after a failed Playwright job
Dry-run first review exact diffs before code changes
Local evidence reports and traces stay on disk
CI-ready GitHub Action comments with patch artifacts
60-second repair walkthrough fail -> analyze -> patch -> green
Animated 60-second demo showing a failed Playwright run moving through Helix analysis, validation, patch generation, and a green rerun
Evidence-backed locator repair 0.85+ confidence gate
Repair Confidence $
0.85+

auto-ready threshold for confident locator replacements

QA Time Recovered T
30-40%

maintenance drag targeted by the MVP painkiller

Patch Posture P
Dry-run

review first with optional automation after validation

Data Boundary L
Local

no lock-in with report, cache, and trace evidence on disk

Core Product Modules

Manage CI flow

Real Report Ingestion

Playwright JSON, failed tests, retries, attachments

Active
Inputs
Reports + traces
Failure map
Spec path + locator
Output
Evidence bundle

DOM-Aware Candidates

Accessibility-first selector generation

Active
Preferred
role/name
Fallback
label/text/test id
Noise guard
ranked confidence

Validation Probe

Static plus live uniqueness checks

Active
Resolve
unique match
Actionable
visible + enabled
Risk
ambiguous flagged

Dry-Run Patch Generator

AST-safe diff suggestions for review

Active
Mode
review-first
Scope
matched locator only
Diagnostics
unsupported cases

GitHub PR Comment

Readable repair summary where developers work

Review
Comment
failure reason
Patch
candidate diff
Artifacts
Markdown + HTML

Report Preview

Confidence, reasoning, and patch output

Active
Helix Heal report preview showing selector candidates, confidence scores, and validation progress

After-Failure Workflow

01

Fail

Playwright CI turns red and saves the report, trace, and source context.

02

Analyze

Helix extracts locator evidence and ranks safer replacements.

03

Validate

Candidate selectors are checked for uniqueness and actionability.

04

Patch

Developers receive a dry-run diff and PR comment with confidence signals.

Real Fixture Walkthrough

Open fixture
Fail

Broken selector

The fixture expects Sign in, but the page renders a safer accessible Log in button.

Analyze

Evidence collected

Helix reads the Playwright JSON report and trace snapshot to find the failed locator context.

Patch

Reviewable diff

The dry-run patch recommends a stable test-id or role/name locator instead of a brittle text selector.

Green

Rerun clean

After applying the reviewed locator, the same Playwright flow can rerun against the changed UI.

npx playwright test
npx helix-heal analyze --report examples/basic-playwright/fixtures/playwright-report.json --trace examples/basic-playwright/fixtures/real-trace --source-root examples/basic-playwright
npx helix-heal patch --dry-run --report examples/basic-playwright/fixtures/playwright-report.json --trace examples/basic-playwright/fixtures/real-trace --source-root examples/basic-playwright
npx playwright test

Operating boundaries

Clear limits, fewer surprises

Helix is built to make locator repair safer, not magical. It flags uncertainty so teams can decide when a human review is the right engineering move.

Unsupported selectors

Complex dynamic locators may receive diagnostics instead of an automatic patch.

Ambiguous DOM

Multiple matching elements lower confidence and require review before a change is trusted.

Live validation

Browser-backed validation needs an installed Playwright browser and a reachable app URL.

Pricing

Start free
Starter $0

Local CLI, dry-run patches, Markdown reports, static dashboard, and community support.

Pro $49/mo

Planned hosted validation, team dashboard, repair analytics, and priority support.

Consultant $199/mo

Planned client workspaces, white-label reporting, branded exports, and delivery support.

Install Flow

Run this after a failed Playwright job

Helix stays easy to evaluate from npm: no hosted account, no vendor lock-in, and no framework migration before the first repair suggestion. 

npx helix-heal analyze \
  --report playwright-report.json \
  --trace test-results \
  --source-root .

npx helix-heal patch --dry-run \
  --report playwright-report.json \
  --trace test-results \
  --source-root .

CI Handoff

Publish a repair plan in the pull request

The action is designed to run only after failure, preserving CI speed while turning breakage into a focused, developer-readable next step. 

- name: Run Playwright
  id: playwright
  run: npx playwright test --reporter=json
  continue-on-error: true

- uses: martjustin/helix-heal@main
  if: steps.playwright.outcome == 'failure'
  with:
    playwright-report: playwright-report.json
    trace: test-results
    source-root: .