Log inSubmit server
Vibe Test — Browser Testing Agent logo

Vibe Test — Browser Testing Agent

Reads application code and tests real browser flows with Playwright through MCP, including coverage, screenshots, reports, and regression diffs.

InstallRepositoryWebsite
Full README Compare alternatives Embed widget Claim this listing

Install Vibe Test — Browser Testing Agent

Pick your client, then add it in one click or paste the config.

STDIONode.js
~/Library/Application Support/Claude/claude_desktop_config.json
claude_desktop_config.json
{
  "mcpServers": {
    "vibe-test-browser-testing-agent": {
      "command": "npx",
      "args": [
        "-y",
        "vibe-testing@latest"
      ]
    }
  }
}

Settings → Developer → Edit Config, merge this into mcpServers, then fully quit and reopen Claude Desktop.

We couldn’t confirm it starts correctly

We ran the install command but it didn't respond within our test window — this can mean a slow first-time install rather than a real problem.

npx -y vibe-testing@latest

No response to initialize.

This is an experimental automated check and can have false negatives — missing environment variables, a slow cold install, etc. It doesn’t necessarily mean something’s wrong. Last checked 27d ago.

Overview

Vibe Test — Browser Testing Agent MCP server reads an application's source code, identifies routes and forms, and runs browser checks through Playwright. Its MCP tools let an AI coding editor inspect source context, authenticate, explore pages, execute scenarios, capture screenshots, generate reports, and compare results with previous runs. Reach for it when tests need real project selectors and route awareness rather than generic browser actions. It runs locally through npx and does not make LLM calls inside the testing tools.

Use cases

  • Scan an application for routes, forms, and coverage gaps
  • Run authenticated browser scenarios against a local app
  • Compare current test results with the previous run
  • Generate HTML reports with screenshots and execution details

Key features

  • Code-derived routes, forms, selectors, and scenarios
  • Playwright browser execution over MCP
  • Authentication and observed API-call results
  • Coverage tracking and prioritized test suggestions
  • Screenshots, HTML reports, and iterative test results
  • Route and snapshot change detection

Tools14

What Vibe Test — Browser Testing Agent lets your agent do. Expand a tool for its parameters.

Self-reported~196 tokens of context

Security & health

What it can reach and how reliably it runs — from our own checks, not the maintainer’s claims.

ActiveLast checked 2d ago

Recent health check succeeded.

20/21 checks healthy over the last 91d

Supply-chain signal

No high-severity advisories surfaced by our automated scan.

Critical 0High 0Medium 0Low 0

Scanned 2d ago via OSV.dev · vibe-testing@latest (npm)

Adoption & maintenance

Factual signals from GitHub, npm, and our automated checks — not a rating.

GitHub stars18Stargazers on the source repository.
npm downloads523Package downloads in the last 30 days.
Last commit2d agoMost recent push to the default branch.
Install checkInconclusiveDidn't respond in our test window — often a slow first install.
Tools exposed14Callable tools this server registers over MCP.
Directory activity5 viewsConfig copies, upvotes, and views on AllMCPs.

How Vibe Test — Browser Testing Agent works

What Vibe Test — Browser Testing Agent does

Vibe Test — Browser Testing Agent MCP server connects an AI coding editor to a local application and a Playwright browser. It derives testing context from the project source instead of relying only on the page currently open in the browser. Code scanning can identify routes, forms, existing coverage, and candidate scenarios, while context lookup returns source details such as field names and selectors.

The server is intended for applications that can be run locally or otherwise reached by a URL. It can test authenticated flows, inspect network activity and browser errors, save screenshots, and produce an HTML report. It also keeps project state under .vibe, allowing later runs to show route changes and differences from the previous test snapshot.

How it works

Initialize the project with the package's init command. The setup process detects available editors, creates MCP and instruction files, attempts to identify the application URL, creates VIBE.md and vibe.config.json, and installs the matching Playwright Chromium build.

A typical workflow starts with project configuration and a codebase scan. The agent can then retrieve feature context, log in if needed, inspect page elements, explore broad behavior, and execute explicit scenarios. Full-test and converge operations combine several of these activities and return a snapshot_diff against the prior run. Codebase scans similarly expose route_changes since the previous scan.

Pass/fail checks are performed with heuristics such as URL changes, toast detection, and API error detection. The browser tools themselves do not call an LLM; the connected editor decides which operations to request and how to interpret the results.

Setup and configuration

The recommended setup is:

Terminal
npx vibe-testing@latest init

For a manual stdio configuration, run npx -y vibe-testing@latest --mcp as the MCP server command. A local development build can instead start dist/mcp-server.js with Node.

Authentication details can be added to the generated VIBE.md. Claude Desktop and other MCPB hosts can use a locally built .mcpb bundle, which requires the repository build steps and a Playwright Chromium prerequisite.

Tools and capabilities

The available MCP operations include:

  • configure for structured project setup.
  • scan_codebase for routes, forms, coverage, generated scenarios, and route changes.
  • get_context for source files and actual selectors.
  • login for authentication results, screenshots, token state, and observed API calls.
  • scan_page_elements, explore_page, and take_screenshot for page inspection and visual evidence.
  • execute_scenario for step logs and screenshots.
  • get_coverage and suggest_tests for finding and prioritizing untested areas.
  • generate_report for an HTML report and summary.
  • run_full_test and run_converge for broader or iterative testing with previous-run diffs.
  • cleanup for closing browsers and releasing resources.

Limitations and notes

Vibe Test — Browser Testing Agent MCP server is code-aware, but it does not replace the connected model's judgment about which scenarios matter. Its verification is heuristic rather than a full assertion framework, using observable browser and API signals described by the project. The server stores run and route snapshots in the project, so comparison results depend on having earlier state to compare against.

The documented runtime requirement is Node.js 20 or newer. The material does not list third-party API keys or required environment variables. Applications that require authentication need suitable test credentials configured through the generated project guidance.

Read the full README View source on GitHub

Reviews

No reviews yet — be the first to share how it worked for you.

Frequently Asked Questions about Vibe Test — Browser Testing Agent

Vibe Test — Browser Testing Agent is an MCP server that connects AI coding editors to a local application and a Playwright browser. Its main tools scan source code, inspect selectors, log in, explore pages, execute scenarios, track coverage, capture screenshots, generate reports, and compare results with previous runs.

Run `npx vibe-testing@latest init` from the project directory. The command detects installed editors, creates MCP configuration and project guidance files, attempts to find the application URL, and installs the matching Playwright Chromium build.

Yes, Vibe Test — Browser Testing Agent works with Cursor. The documented setup can register it globally in `~/.cursor/mcp.json` or for one project in `.cursor/mcp.json`, using `npx -y vibe-testing@latest --mcp` as the server command.

It compares the current run with the previous stored snapshot. Each run writes `.vibe/run-snapshot.json`, and the full-test and converge tools return a `snapshot_diff` showing changes such as fixes and regressions; codebase scans also report route changes.

No, the testing tools do not require an API key for their own LLM processing because they make no LLM calls. The connected editor's model chooses testing actions, while the server performs browser operations and heuristic checks locally.

Related MCP servers

All in Browser Automation All alternatives
Add a badge to your READMEShow it’s listed — live SVG, Markdown or HTML.

AllMCPs Directory Badge

Full Badge Customizer

Showcase your server listing on GitHub or your project documentation. Embed this dynamic SVG badge to highlight official listing status and live engagement.

Badge Style:
Live Dynamic SVG PreviewVibe Test — Browser Testing Agent AllMCPs Directory Badge
Markdown (GitHub README)
[![AllMCPs](https://allmcps.com/api/badge/vibe-test-browser-testing-agent?style=directory)](https://allmcps.com/mcp/vibe-test-browser-testing-agent)
HTML Embed
<a href="https://allmcps.com/mcp/vibe-test-browser-testing-agent"><img src="https://allmcps.com/api/badge/vibe-test-browser-testing-agent?style=directory" alt="Vibe Test — Browser Testing Agent on AllMCPs" /></a>