- Category
- Developer Tools
- Rank
- No. 1459Tools index
Previous survey · No. 1467 ·
- Listed in
- #2 Generate and improve tests
- Pricing
- Open Source
- Type
- TOOL
- Builder
- browser-use
- GitHub
- 603 stars
- Date
About
QA automation tool from the Browser Use team for running agent-driven test flows against web apps.
What it does
You write browser journeys and success criteria in plain English, then organize them into reusable suites. BrowserUse executes the interactions, while qa-use tracks progress, decides pass or fail, and presents reports, screenshots, recordings, schedules, and failure notifications through a web interface.
Why it's ranked here
The project offers more than a thin prompt runner. It combines suite management, parallel execution, persistence, scheduling, imports, and run evidence in one self-hosted application. The trade is dependence on a paid external execution service, plus coarse scheduling and validation built around agent judgment rather than deterministic assertions.
What's good
Tests remain readable because steps and expected outcomes use ordinary language. Suites can run together, execute tests in parallel, and retain run status plus links to live and shared recordings. JSON imports receive schema validation and a size limit. Database transactions prevent partially imported suites.
Tradeoffs
Running it requires Docker, PostgreSQL, and a BrowserUse API key tied to an active subscription for task execution. Scheduling supports only hourly, daily, or disabled states in the shown data model. Status updates refresh the page every two seconds. Natural-language evaluation may not provide the precision of code assertions.
How to use it well
Use it for teams that want readable browser checks, scheduled smoke suites, and visible execution evidence without authoring selector-heavy scripts. Start with narrow journeys and measurable success criteria, then use failure emails for recurring checks. Keep deterministic unit, integration, accessibility, performance, and fine-grained browser tests alongside it.
Technical notes+
package.json defines a private Next.js 15 TypeScript application with build, development, start, lint, type-checking, and OpenAPI generation scripts. src/lib/api/client.ts creates a typed openapi-fetch client for the BrowserUse API using BROWSER_USE_API_KEY, while src/lib/api/v1.d.ts contains generated endpoint types. src/lib/db/db.ts connects Drizzle ORM to PostgreSQL through DATABASE_URL, and src/lib/db/schema.ts models suites, ordered steps, suite runs, test runs, statuses, scheduling, notification addresses, and recording links. src/app/actions.ts validates creation and JSON imports with Zod, rejects empty or over-10MB files, and imports each suite inside a transaction. src/components/Polling.tsx refreshes active run views every two seconds by default.
Observed
- License
- MIT
- Primary language
- TypeScript and TSX
- Install surface
- Self-hosted with Docker and Docker Compose
- Interface
- Next.js web application
- Persistence
- PostgreSQL through Drizzle ORM
- Required service
- BrowserUse API key
- Optional integration
- Resend email notifications
Read from README.md, package.json, src/app/page.tsx, src/app/types.ts, src/lib/types.ts, src/lib/utils.ts, src/app/loader.ts, src/app/actions.ts, src/app/layout.tsx, src/hooks/useMounted.tsx, src/components/Polling.tsx, src/lib/db/db.ts, src/lib/api/v1.d.ts, src/lib/db/schema.ts, src/lib/api/client.ts.
What it can do
Run automated test flows against web applications
Web application URL and test scenarios → Test execution results and reports
Execute agent-driven browser interactions
Test instructions and target web pages → Automated browser actions and responses
Validate web application functionality
Application endpoints and expected behaviors → Pass/fail test results and validation reports
Generate QA test reports
Test execution data and results → Formatted test reports and metrics
Simulate user interactions with web interfaces
User workflow definitions and web application → Automated click, type, and navigation actions
Tags
Tech Stack
Comments (0)
No comments yet
Editorially curated, with community endorsements as a secondary signal. Corrections welcome.