About Server Watch
Server Watch wraps a running command, captures its stdout and stderr, and serves the output through an HTTP-based MCP server. It is intended for development workflows that need access to process logs or build output.
Key features
- 01Wraps and runs a command
- 02Captures stdout and stderr in real time
- 03Keeps the last 5000 log entries in memory
- 04Retrieves logs with stream filtering
- 05Searches logs case-insensitively
Use cases
- 01Inspect a development server’s recent output
- 02Find errors in build logs
- 03Monitor output from a running command
MCP support
It exposes process output through an HTTP-based MCP server, with `get_logs` and `search_logs` tools for connected MCP clients.
Configuration
Remote servers need no local install. Pick your client, then copy the config or install in one click.
http://localhost:6280/sseclaude mcp add --transport sse server-watch http://localhost:6280/sse{"mcpServers":{"server-watch":{"type":"sse","url":"http://localhost:6280/sse"}}}Open in Cursor{"servers":{"server-watch":{"type":"sse","url":"http://localhost:6280/sse"}}}Open in VS Code{"mcpServers":{"server-watch":{"type":"sse","url":"http://localhost:6280/sse"}}}{"mcpServers":{"server-watch":{"serverUrl":"http://localhost:6280/sse"}}}Run commands in a terminal; paste JSON into the client’s MCP config and restart. Servers that need sign-in will prompt on first connect.
Tools
Details
From the project READMEserver-watch-mcp
A straight-forward CLI wrapper that monitors and captures output from any running command and exposes it as an MCP server. Perfect for development workflows where you need to access server logs, build output, or any process output.
Features
- HTTP-based MCP server - Run as a standalone service that clients can connect to
- Real-time log capture - Captures stdout and stderr from any command
- Circular buffer storage - Maintains last 5000 log entries in memory
- Two MCP tools:
get_logs- Retrieve recent logs with optional filtering by stream typesearch_logs- Search through logs with case-insensitive substring matching
Installation
npm install -g server-watch-mcp
# or
pnpm add -g server-watch-mcp
# or in your project
pnpm add -D server-watch-mcp
Usage
1. Add the MCP server to Claude Code
claude mcp add -s user -t sse server-watch-mcp http://localhost:6280/sse
Which should configure it like this:
{
"mcpServers": {
"server-watch-mcp": {
"type": "sse", // Even though SSE is deprecated, Claude Code only supports SSE
"url": "http://localhost:6280/sse"
}
}
}
2. Start the MCP server with your command
option 1: wrap the command when calling it
# Monitor a development server
server-watch-mcp npm run dev
# Monitor a build process
server-watch-mcp npm run build:watch
# Monitor any command
server-watch-mcp python app.py
option 2: wrap the command in your package.json (or script file)
// In your package.json
{
"scripts": {
"dev": "server-watch-mcp next dev --turbo",
}
}
The server will:
- Start an HTTP server on port
6280(orSERVER_WATCH_MCP_PORTenvironment variable) - Execute your command as a child process
- Capture all output from the command
- Continue running even if the child process exits
3. Use the tools in Claude
Once connected, you can use these tools:
- Get recent logs: "Show me the last 50 logs from stderr"
- Search logs: "Search for any errors in the logs"
- Monitor output: "What's happening with my dev server?"
Environment Variables
SERVER_WATCH_MCP_PORT - Override the default port (6280)
SERVER_WATCH_MCP_PORT=6281 server-watch-mcp npm run dev
Development
# Install dependencies
pnpm install
# Build
pnpm run build
# Run tests
pnpm test
# Run in development
pnpm run dev
License
ISC
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.