home / mcp / mcp analyzer mcp server
An MCP server to read MCP logs to debug directly inside the client
Configuration
View docs{
"mcpServers": {
"klara-research-mcp-analyzer": {
"command": "node",
"args": [
"/absolute/path/MCP-Analyzer/build"
]
}
}
}You can analyze and debug MCP logs with a specialized MCP server that reads log files from common locations, filters entries, and presents findings in a paginated, manageable format. It integrates with Claude Desktop for seamless log inspection during MCP workflows, helping you troubleshoot connections and understand how MCP tools interact with your environment.
Install and run the MCP Log Reader on your machine, then connect it to your MCP client to start inspecting logs. You can search across log files, filter results by text, and browse large collections efficiently with pagination. When you need to investigate a specific issue, set a filter (for example, containing the word error) and navigate pages to find the root cause without loading entire files.
# Prerequisites
- Node.js installed on your system (12.x+ recommended)
- npm (comes with Node.js) or yarn
# Quick install via MCP-friendly client (if you have a package that exposes MCP tools)
# This step may vary based on your setup; use the official install flow provided by the MCP tool you’re using
npx -y @smithery/cli install @klara-research/MCP-Analyzer --client claude
# Manual installation
# Clone the MCP-Analyzer repository
git clone https://github.com/klara-research/MCP-Analyzer.git
cd MCP-Analyzer
# Install dependencies
npm i
# Build the TypeScript sources
npx tscTo connect the reader to Claude Desktop, configure Claude to launch the local MCP server using the runtime command shown below. This enables Claude to invoke the log reader and retrieve log data for analysis.
Configuration snippets show how to wire the log reader into Claude Desktop. Use the exact paths provided in your environment when configuring the integration.
Notes: - The reader automatically detects your operating system and locates the appropriate log directory. - It sorts MCP log files by modification time with the newest first. - Large files are processed with size limits to prevent overwhelming responses. - Results include pagination details so you can navigate through logs efficiently.
If you need to adjust how many lines are read per file or how many files are shown per page, modify the available parameters accordingly.
- readLogs: Retrieve log entries from MCP log files across supported platforms. Return entries in a structured format suitable for quick review.
- filterLogs: Apply a case-insensitive text filter to log entries to quickly surface relevant events.
- paginateLogs: Provide paginated access to logs so you can browse large collections without loading everything at once.
- truncateLargeFiles: Manage very large log files by truncating or streaming data to maintain responsive responses.
- claudeBridge: Integrates log data with Claude Desktop to support conversational analysis and targeted troubleshooting.
Ensure your log data access is restricted to trusted clients. If you encounter connection issues with Claude, verify that the MCP server is reachable from your Claude environment and that the correct command path to the build output is used.
This MCP server is designed to work across macOS, Windows, and Linux. It reads standard log paths and provides a consistent interface for searching and reviewing MCP activity.
Retrieve log entries from MCP log files across supported platforms and present them in a structured format for easy review.
Apply a case-insensitive text filter to log entries to surface relevant events quickly.
Provide paginated access to logs so you can browse large collections without loading everything at once.
Handle very large log files by truncating content or streaming data to maintain responsive responses.
Integrate log data with Claude Desktop to enable conversational analysis and targeted troubleshooting.