Know what your PR
breaks in QA
before CI does
A dev changes a selector or a route — and your Playwright/Cypress/Selenium suite goes red an hour later. Testward reads each PR and flags the exact automation it will break, even when your tests live in a separate repo.
No more "what changed, what broke, what do I update?" · Free on public repos · No credit card
The comment your QA team wishes every PR had
🟠 Testward — Medium Risk (64%)
Summary:
Renames the login button's test id and changes the post-login redirect
from /dashboard to /home.
🚨 Automation this PR may break:
acme/web-e2e→tests/login.spec.ts— clicks[data-testid=login-btn](renamed tosignin-btn) and asserts the/dashboardredirect that changed to/home
Why this PR is notable:
- Changes user-facing selectors that automation depends on
- Alters a navigation route asserted on in E2E specs
— Testward · AI analysis · was this helpful?
What you get on every PR
Cross-repo automation impact
The big one. When a PR changes a selector, route, or endpoint, Testward finds the Playwright/Cypress/Selenium specs that depend on it — even in a separate automation repo — and names exactly what'll break. How to link your repo →
Tests that may need updating
For tests in the same repo: Testward reads the covering test and tells you which ones the change invalidates — e.g. "asserts on the old token shape." What to fix, not just run.
Which tests cover the change
Zero setup: maps changed files to the tests that cover them. Connect your CI and it ranks them by real failure history too (Pro).
Risk score + plain-English summary
One comment, not forty. A calibrated risk score and a one-paragraph summary of what the PR actually does — so reviewers know what they're looking at before the Files tab.
Secret leak detection Pro
Catches AWS keys, Stripe tokens, OpenAI keys, JWTs in the diff. Before they land on main.
Reviewers + Slack alerts Pro / Team
Suggested reviewers from git history, and a Slack ping when a high-risk PR lands — without writing a GitHub Action.
The math QA leads already know
Every time a PR silently breaks automation, someone on your team loses an afternoon to "what changed, what broke, what do I update." Testward moves that to a 10-second read on the PR.
Breakages caught at PR-open time, not after a 20-minute pipeline goes red.
The spec, the selector, and the cause are named for you. No bisecting, no guessing.
Less than 30 minutes of one engineer's time. It pays for itself the first PR it catches.
Pricing that doesn't punish open source
Public repos are free forever. Private repos start at $19/mo.
For open-source and trying us out.
- ✓ Unlimited public repos
- ✓ 50 private-repo PRs/month
- ✓ Risk score + AI summary
- — Reviewer suggestions
- — Secret scanning
For teams shipping private code daily.
- ✓ Everything in Free
- ✓ Unlimited private-repo PRs
- ✓ Suggested reviewers
- ✓ Secret leak detection
- ✓ Email support
Install free, upgrade in one click from your first PR comment
Per org. For orgs that want the full stack.
- ✓ Everything in Pro
- ✓ Slack alerts on high-risk PRs
- ✓ Priority LLM (faster comments)
- ✓ Team analytics dashboard
- ✓ Priority support
Install free, upgrade from your first PR comment
Common questions
Does Testward read my code to train an AI?
No. Diffs are sent to the LLM per-PR to generate the comment, and then discarded. We do not train on your code. On Pro/Team you can bring your own API key.
My automation lives in a different repo from the app. Does that work?
Yes — that's the whole point. Install Testward on both repos and add a one-line .testward.yml to the app repo listing your automation repos. See the setup →
How is this different from CodeRabbit or Greptile?
They're general AI reviewers that write line-by-line nitpicks on the code itself. Testward is built for QA/automation teams: it connects an app PR to your test automation — including in a separate repo — and tells you what will break. Full comparison →
What data do you store?
Installation metadata, PR numbers, the comment we posted, and (if you connect CI) test pass/fail outcomes. We do not store your source code or diffs — they're sent to the LLM for a single analysis and discarded. We don't train on your code.
Which test frameworks does it support?
It's framework-agnostic — it matches the selectors, routes, accessibility labels, and visible text your specs depend on, not a specific tool's API. That covers Playwright, Cypress, Selenium, WebdriverIO, Puppeteer, TestCafe, Appium, Testing Library, Cucumber, and Robot Framework, across JS/TS, Python, Ruby, Java, C#, Kotlin, Swift, Go and PHP. CI history ingestion (Pro) accepts JUnit XML, which virtually every framework emits.
Ship faster. Break less.
One click, one comment per PR, one checkout when you're ready.
Install Testward on GitHub