mcp-shark
Wireshark-like forensic analysis for Model Context Protocol communications Capture, inspect, and investigate all HTTP requests and responses between your IDE and MCP servers
claude mcp add --transport stdio mcp-shark-mcp-shark npx -y @mcp-shark/mcp-shark --open
How to use
MCP Shark is a monitoring and aggregation tool that connects to multiple MCP servers and presents a unified interface for observing MCP traffic, running analyses, and testing endpoints. It provides a real-time, Wireshark-like view of MCP communications, an interactive playground for prompts and resources, and AI-assisted security scans to identify risks across connected servers. The UI also automates IDE integrations by detecting configuration for popular editors, and exposes comprehensive API documentation with interactive testing via Swagger/OpenAPI.
To use MCP Shark, start the server (via npx or a global install) and open the web UI. From there you can add and manage multiple MCP servers, inspect live traffic, run local analyses, and leverage the AI-powered scanning features to surface potential security concerns. The interactive playground enables you to experiment with prompts and resources directly within the UI, while the API docs provide endpoints and WebSocket protocols for integration with your own tooling.
How to install
Prerequisites:
- Node.js 20.0.0 or higher
- npm (comes with Node.js) or pnpm/yarn if you prefer
Option 1: Run with npx (no installation required)
npx @mcp-shark/mcp-shark --open
This will download the package on demand and launch the MCP Shark UI in your browser.
Option 2: Install globally via npm
npm install -g @mcp-shark/mcp-shark
mcp-shark --open
Option 3: Install as a project dependency (for local development)
mkdir my-mcp-project
cd my-mcp-project
npm init -y
npm install @mcp-shark/mcp-shark
npx mcp-shark --open
Notes:
- Ensure your environment has network access to fetch the package from npm registry.
- If you run behind a proxy, configure npm accordingly and retry.
- The UI will host on http://localhost:9853 by default unless configured otherwise.
Additional notes
- The server supports adding multiple MCP endpoints; use the UI to configure each MCP server URL and credentials if required.
- The API documentation (/api-docs) is interactive once the server is running.
- For environment-specific setups, consider configuring environment variables if provided by MCP Shark (check docs/configuration.md).
- If you encounter issues starting via npx, try clearing npm cache or reinstalling Node.js LTS version and try again.
- Check the license terms in the project’s LICENSE file; this is a Non-Commercial license.
- Review the Quick Start and Documentation sections in the UI for feature-specific guidance.
Related MCP Servers
MCP-Kali
MCP configuration to connect AI agent to a Linux machine.
MCPSpy
MCP Monitoring with eBPF
mcp-hub
A centralized manager for Model Context Protocol (MCP) servers with dynamic server management and monitoring
asterisk
Asterisk Model Context Protocol (MCP) server.
mdk
Model Context Protocol (MCP) server for AI-assisted development ("vibe coding") of MDK applications.
architect
A powerful, self-extending MCP server for dynamic AI tool orchestration. Features sandboxed JS execution, capability-based security, automated rate limiting, marketplace integration, and a built-in monitoring dashboard. Built for the Model Context Protocol (MCP).