Find bugs before
your users do.
Tests anything that runs in a browser
One agent. Two modes.
Vera is the QA engineer. She works black-box from a URL today, and code-aware from your repo next — and you reach her from the dashboard, your CI pipeline, or your editor.
Vera Explore
Black-box · URL or build
Drop a URL. Vera crawls page by page — login, checkout, onboarding — like a senior QA engineer, and reports only verified bugs in minutes. No source code required.
- Goal-driven browser navigation
- Explore → Plan → Review → Judge → Report
- Auth-aware — credentials, cookies, OAuth
Vera Inspect
White-box · repo or code
Point Vera at a repo and she reads the source — pairing static analysis with live execution to pin each bug to a file:line. The coverage black-box can't reach.
- Code-grounded test planning
- Static defects + reproduced at runtime
- Reuses the same Judge · Report
Vera Explore Session
Run from the Verdict Web dashboard — sessions, reports, credits.
verdict-app.net
Vera Explore CLI
Quality gate in CI — fail the build on critical bugs.
npm i -g @verdict-npm/cli
Vera Explore MCP
Editor-native — call Vera from Claude Code or Cursor.
npx -y @verdict-npm/mcp
A 5-stage pipeline.
Verified bugs only.
This is the exact pipeline you'll watch run in the dashboard. Each stage has a distinct role — and you can step in at Review to shape the test plan before Vera judges it.
Explore
Vera navigates autonomouslyDrop a URL. Vera uses an AI browser agent to explore every flow — login, checkout, onboarding — goal-first, without a script.
Plan
Test cases draftedVera turns what it discovered into a structured test plan — concrete test cases with expected outcomes.
Review
Optional human checkpointEdit, add, or drop test cases before judging — or skip the checkpoint and let Vera run end-to-end.
Judge
AI QA evaluator decidesA separate QA Judge LLM evaluates whether each anomaly is a real bug or expected behavior — so noise never reaches your report.
Report
Evidence, steps, severityA clean report with reproduction steps, screenshots, and severity scores — ready to drop into your tracker.
Not another test script.
Vera thinks like a QA engineer — building hypotheses, constructing edge cases, proving bugs are real.
Goal-driven exploration
No scripts to write or maintain. Vera sets goals — “complete checkout” — and navigates autonomously, finding the paths a human tester would.
Auth-aware testing
Real bugs live behind the login wall. Hand Vera credentials, cookies, or tokens and she tests your app the way a signed-in user experiences it.
A quality gate in CI
One command in any pipeline. If Vera confirms a critical bug on staging, the build fails — before the bug reaches production.
$ verdict explore staging.acme.dev --fail-on critical
✗ 1 critical finding — exiting 1, build failed
If it's software, Vera can test it.
Web is live today. Every other surface is on the roadmap, built on the same pipeline through platform-specific adapters.
Simple credit-based pricing
1 credit = 1 browser step = $0.02. Pay for what you actually use — quick runs cost less, deep explorations cost more.
- ✓ Web testing
- ✓ Basic sessions
- ✓ Community support
- ✓ Web + API testing
- ✓ Email support
- ✓ Web + API + Mobile
- ✓ Priority support
- ✓ Team seats (3)
- + Deep explore
- ✓ All platforms
- ✓ SSO / SAML
- ✓ Dedicated CSM
- ✓ SLA 99.9%
1 credit = 1 browser step = $0.02. Vision analysis is included on every step at no extra cost. A minimal session costs ~10 cr; a typical session 30–100 cr.
Deep explore (Pro+) lets Vera crawl more pages per session for broader coverage. Vision analysis is included on every step — never billed as an add-on.
Yes. Bundles: 50 cr/$1, 100 cr/$1.90 (Free+) · 250 cr/$4.50, 500 cr/$8 (Starter+) · 1,000 cr/$15, 2,000 cr/$28 (Pro+). Purchased credits never expire.
Enterprise: custom credit volumes, on-premise/VPC deployment, white-glove onboarding. Talk to us →
Start shipping with
certainty today.
Paste a URL and watch Vera explore it live — free, no account needed.