Ranger is a QA platform that automatically verifies features built by AI coding agents in a real browser. Running a single command, ranger go, spins up a browser agent that drives the screen like a human would and leaves behind screenshots, video recordings, and Playwright traces as evidence. Results are gathered into a view that resembles a GitHub pull request review, and when a defect is found it can be handed straight back to the coding agent for automatic fixing and re-verification. Approved scenarios convert into permanent E2E tests with one click, and AI also handles root-cause investigation when a test breaks. Ranger works with Claude Code, Cursor, and Codex, and launched out of the US in 2025.
Key Features
- Autonomous verification from one command: Running
ranger golaunches a browser in the cloud, walks the specified user flow to the end, and returns a verdict. No test code has to be written up front - Verification runs while the agent builds: Checking starts as soon as the coding agent finishes a feature, and any defects are reported back to the agent. The agent fixes and re-verifies, so the loop keeps moving without a human in the middle
- Evidence is always retained: Every verification carries screenshots, video, and a Playwright trace, so an agent’s “it works” claim can be confirmed against the actual screen
- PR-review-style interface: Results are grouped per change so the team can comment and sign off before deploying, matching the review flow already used on GitHub
- One-click E2E tests: Approved verification scenarios can be added directly to a permanent E2E suite. AI also takes on root-cause investigation when a test fails, which makes broken tests less likely to be left unattended
- Massive parallel execution: The infrastructure scales to hundreds of simultaneous browser instances, so verification time does not balloon as the suite grows
Pricing
| Plan | Price | What’s included |
|---|---|---|
| Free | $0 | Up to 5 reviews, self-serve install |
| Growth | $50/month per user | Unlimited reviews, self-serve install, support via Slack or email |
| Enterprise | Usage-based (contact sales) | Custom onboarding, support via Slack or email |
Pricing is as of August 2026. Check the official site for the latest pricing.
Pros & Cons
✅ Pros
- You can start without writing test code, so even a project with no E2E coverage can run verification from day one
- Screenshots, recordings, and traces are always retained, letting you visually confirm an agent’s completion claim
- The loop from defect hand-back to re-verification is closed, so no human has to mediate the build-verify-fix round trip
- Verification scenarios become E2E tests, so the work accumulates as an asset instead of being thrown away
- The free plan’s five reviews let you try it on your own product before committing
⚠️ Cons
- The focus is browser interaction with web apps; native apps and API-only testing are out of scope
- The Growth plan is priced per user, so costs add up as more reviewers join
- Some capabilities such as CI integration and background agents are described as beta, so maturity varies
- Before delegating verification, the target environment has to be reachable from a browser (a preview environment, working sign-in, and so on)
- It is a relatively new service, and Japanese-language case studies and write-ups are still scarce
Comparison with Similar Services
| Criteria | Ranger | QA Wolf | Playwright (self-managed) | Momentic |
|---|---|---|---|---|
| Positioning | Automated verification for coding agents | Managed E2E testing service | Test automation framework | AI-assisted test creation and maintenance |
| Test creation | Automatic, from natural-language instructions | Built and maintained by the vendor’s team | Written in code by developers | Natural language plus low-code |
| Agent integration | Claude Code / Cursor / Codex | Limited | Must be wired up yourself | Partial |
| Evidence | Screenshots, recordings, traces | Reports and recordings | Trace viewer | Recordings and logs |
| Pricing | Free tier; from $50/month | Contact sales | Free (OSS; runtime costs separate) | Contact sales |
Who Is It For
- Developers who have Claude Code or Cursor writing their code, but find it a burden to open a browser and check the result every time
- Small teams that want E2E tests but keep shipping features before ever getting started on them
- Reviewers who want an agent’s “implementation complete” report backed by evidence
- Teams losing time to root-cause investigation every time a test breaks, watching their suite slide into irrelevance
- Product teams that want fewer production incidents without slowing release velocity
Summary
Ranger fills the gap that opens up when AI writes the code: who checks the result. One command sends a browser agent through the real screen, evidence is retained, and defects go back to the agent. Because that loop is closed, far less human mediation is needed between implementation and verification. The fact that approved verifications survive as E2E tests is also pragmatic for teams that keep postponing test coverage. Start with the free plan’s five reviews and see how far the agent can walk through your own product.