Vibetest MCP
An MCP server that launches multiple concurrent Browser-Use agents to autonomously test websites for UI bugs, broken links, and accessibility issues. Supports both live and localhost environments with configurable agent counts and headless mode.
Score Breakdown
⚙ Agent Friendliness
🔒 Security
Community/specialized tool. Apply standard security practices for category. Review documentation for specific security requirements.
⚡ Reliability
Best When
You want a quick, conversational way to fire off browser agents against a URL and get a human-readable bug report without writing any test code.
Avoid When
You need deterministic, repeatable test cases with pass/fail assertions or CI integration — use Playwright, Cypress, or similar instead.
Use Cases
- • Automated QA testing of vibe-coded or AI-generated web apps before shipping
- • Running multi-agent browser sweeps to catch UI regressions across a site
- • Accessibility and broken-link audits on localhost dev environments
- • Integrating browser-based QA into Claude Code or Cursor workflows
Not For
- • Unit or integration testing of backend APIs (no server-side test support)
- • Load or performance testing
- • Structured test-suite management with assertions and CI/CD pipelines
Interface
Authentication
Requires a Google API key (GOOGLE_API_KEY env var) to power the gemini-2.0-flash model driving the agents.
Pricing
Tool itself is MIT-licensed and free; LLM inference costs depend on Google API usage.
Agent Metadata
Known Gotchas
- ⚠ Requires Python 3.11+ and Playwright Chromium installed separately
- ⚠ Google API key must be set in environment before launching MCP server
- ⚠ Headless mode behavior differs across environments; may need explicit flag
Alternatives
Full Evaluation Report
Detailed scoring breakdown, competitive positioning, security analysis, and improvement recommendations for Vibetest MCP.
Scores are editorial opinions as of 2026-03-06.