Vibe Test — Browser Testing Agent
Code-aware Playwright browser testing for AI coding editors, via MCP or CLI
- Pricing
- Open source
- Platforms
- CLI, API, Web
- Runs
- Locally
- Version
- 0.5.0
- Listing updated
- today
Similar MCP servers
About Vibe Test — Browser Testing Agent
Vibe Test — Browser Testing Agent is an open-source MCP server for browser automation. It runs locally on your machine and is started by your AI client.
Vibe Test reads your project's source so its browser tests use real routes and form field names instead of guessed selectors. It then runs those tests in Playwright Chromium and keeps memory of earlier runs, so each new run can show what got fixed or broke since the last one.
It is delivered as an MCP server exposing 14 testing tools to editors such as Claude Code, Cursor, Windsurf, VS Code Copilot and Roo Code, and also as a standalone CLI. The README says the tool makes no LLM calls itself; the editor's model chooses what to test while the tool handles browsing and heuristic pass/fail checks.
The README describes it as aimed at developers who build with AI coding agents and want a repeatable regression check on their app. Source code is optional: pointed at a bare URL, it runs a black-box pass instead.
Install
Run in a terminal
claude mcp add vibe-testing -- npx -y vibe-testingAdd to claude_desktop_config.json
{
"mcpServers": {
"vibe-testing": {
"command": "npx",
"args": [
"-y",
"vibe-testing"
]
}
}
}Add to ~/.cursor/mcp.json
{
"mcpServers": {
"vibe-testing": {
"command": "npx",
"args": [
"-y",
"vibe-testing"
]
}
}
}Add to .vscode/mcp.json
{
"servers": {
"vibe-testing": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"vibe-testing"
]
}
}
}Add to ~/.codeium/windsurf/mcp_config.json
{
"mcpServers": {
"vibe-testing": {
"command": "npx",
"args": [
"-y",
"vibe-testing"
]
}
}
}Add to your client's MCP settings (most use this shape)
{
"mcpServers": {
"vibe-testing": {
"command": "npx",
"args": [
"-y",
"vibe-testing"
]
}
}
}Using several servers? Build one config file with the MCP config generator.
Tools this server provides (14)
Taken from the project documentation; the running server may differ.
- configure
- Detects framework, server, and authentication for a new project.
- scan_codebase
- Reads source code and finds routes, forms, tests and gaps.
- get_context
- Returns source files for a feature.
- login
- Authenticates in a real browser.
- scan_page_elements
- Lists interactive elements on a page.
- explore_page
- Interacts with a page and reports problems.
- execute_scenario
- Runs specific test steps.
- get_coverage
- Shows coverage map and untested routes.
- suggest_tests
- Finds coverage gaps and suggests scenarios.
- take_screenshot
- Captures a screenshot of a URL.
- generate_report
- Builds an HTML report.
- run_full_test
- Scans, executes, explores and reports in one run.
- run_converge
- Iterates testing until thresholds are met.
- cleanup
- Closes browsers and frees resources.
What it does
- Source-aware scenario generation
- Scans routes, forms, and existing test suites to derive scenarios that use the app's actual field names.
- Browser exploration
- Clicks through pages in Playwright and reports server errors, console errors, and broken values.
- Run-to-run regression diff
- Stores a snapshot after each run and lists newly failing and newly passing routes.
- Configurable safety limits
- Lets you block destructive actions and control whether exploration may change app data.
- HTML reports with screenshots
- Generates a report with screenshots of each step and a summary of failed scenarios.
Good for
- Check a web app for regressions after an AI-assisted code change
- Smoke-test a staging environment from the command line
- Find broken routes and console errors on a project without writing test cases
- Verify a specific bug fix through a real browser session
Pricing
Open source
MIT licensed, free to install via npm.
Vibe Test — Browser Testing Agent FAQ
Does it call an LLM internally?
No. The README says verification is heuristic and the editor's model decides what to test.
Which editors are supported?
The README lists Claude Code, Cursor, Windsurf, VS Code Copilot and Roo Code, plus Codex and Cline configuration examples.
Does it need source code?
No. Given only a URL, it tests the app black-box, with lighter coverage than when source is available.
Is Vibe Test — Browser Testing Agent free?
Yes. Vibe Test — Browser Testing Agent is open source, so you can use and self-host it for free.
How do I add Vibe Test — Browser Testing Agent to Claude Code?
Run this command in your terminal: claude mcp add vibe-testing -- npx -y vibe-testing
What are the best alternatives to Vibe Test — Browser Testing Agent?
The closest alternatives to Vibe Test — Browser Testing Agent are VibeView, Vibe Shipped, VibeKit, Playwright MCP and Momentic.
Where this comes from
- Description: written from the project's README and its MCP Registry entry.
- Install steps: from the official MCP Registry entry.