Browser Instrumentation

An MCP server for browser instrumentation that prioritizes observation over action using Playwright, delivering visibility into browser events.

python

0

GitHub Stars

python

Language

5 months ago

First Indexed

2 months ago

Catalog Refreshed

Documentation & install

Readme and setup notes from the catalogue, plus a client-ready config you can copy for your MCP host.

Installation

Add the following to your MCP client configuration file.

Configuration

View docs

{
  "mcpServers": {
    "kvfxkaido-browser-instrumentation-mcp": {
      "command": "python",
      "args": [
        "-m",
        "browser_instrumentation_mcp.server"
      ]
    }
  }
}

This MCP server provides browser instrumentation using Playwright, focusing on observing browser behavior and events. It exposes a set of observation tools that let you inspect navigation, DOM, console output, network activity, and more, while escalating actions only when you deliberately enable them and provide justification.

How to use

You interact with this MCP server through an MCP client to create sessions, observe browser activity, and optionally perform limited actions with escalation. Start with safe, observation-only workflows to explore how pages render and respond, then escalate only when you need to perform an action and you understand the risks.

Typical usage patterns: create a session, navigate to a page, capture screenshots or DOM content, and review the event log to audit what happened. If you need to test an interaction, escalate the session with a clear justification, then perform the action and review the observed changes and logs.

How to install

# Prerequisites
python3.11+
playwright install chromium

# Install from source
git clone https://github.com/kvfxkaido/browser-instrumentation-mcp.git
cd browser-instrumentation-mcp
pip install -e .

Configuration

To run the server locally, use the standard Python module invocation. You will run the server in the command line, and connect your MCP client to it using the defined client protocol.

{
  "mcpServers": {
    "browser": {
      "command": "python",
      "args": ["-m", "browser_instrumentation_mcp.server"]
    }
  }
}

Development and testing

Development setup includes installing dependencies, setting up Playwright browsers, and running unit and integration tests. You can run unit tests quickly without requiring a real browser, and run integration tests when you have a Chromium browser available.

# Clone and install in editable mode with dev dependencies
git clone https://github.com/kvfxkaido/browser-instrumentation-mcp.git
cd browser-instrumentation-mcp
pip install -e .[dev]
playwright install chromium

# Run the server directly
python -m browser_instrumentation_mcp.server

Notes and best practices

Use observation tools first to minimize risk and side effects. - Escalate only when you need to perform an action and have a justification. - Actions do not report success or failure; they report observed changes and a confidence level. - Audit trails are kept in an event log per session for transparency.

Available tools

Session management

Create, list, and destroy sessions; escalate to enable actions within a session.

INSPECT tools

Navigate, screenshot, DOM, text, console, network, and events inspection without side effects.

ACT tools

Click, type, and execute scripts within an escalated session; observe results rather than assuming success.

Event log

Maintain an append-only log of all session events for audit and review.