MCP Telemetry
About
Observability helps. This MCP server adds tracing to all your conversations on Claude (or suitable MCP client) so that you can trace, understand, debug and report on your all your interactions.
Details
- Author
- xprilion
- Downloads
- 479
- Categories
- Developer Tools
Jump to
- Start tracing sessions with custom identifiers
- Log user inputs, LLM responses, actions, and tool calls
- Seamless integration with Weights & Biases Weave
- Real-time monitoring of conversation flows
- Export and share conversation analytics
Setting up with Highlight
This MCP is not yet compatible with Highlight’s one-click setup. However, you can still use it with Highlight by following these steps:
- Download and install Highlight from highlightai.com/download
- Navigate to the plugins tab and select "Add Custom Plugin"
-
Configure the plugin with the settings below
Plugin Name
MCP TelemetryCommand (node, npx, python, etc.)Please refer to the README for specific instructions on how to obtain API keys or other required environment variables.
- Enable "Start Automatically" if you want the plugin to start when Highlight launches
From the repository
Install by adding the provided JSON configuration to your Claude desktop config, setting the WANDB_API_KEY environment variable. The server starts automatically with Claude and logs conversation data, viewable in the Weights & Biases dashboard.
Claude Desktop / Cursor
Paste into your MCP client config file to install this server.
{
"mcpServers": {
"mcp telemetry": {
"MCP Telemetry": {
"command": "uv",
"args": [
"run",
"--with",
"mcp[cli]",
"--with",
"weave",
"mcp",
"run",
"~/mcp-telemetry/server.py"
],
"env": {
"WANDB_API_KEY": ""
}
}
}
}
}McpServers
{
"MCP Telemetry": {
"command": "uv",
"args": [
"run",
"--with",
"mcp[cli]",
"--with",
"weave",
"mcp",
"run",
"~/mcp-telemetry/server.py"
],
"env": {
"WANDB_API_KEY": ""
}
}
}MCP Telemetry
<div align="center">
<strong>A Model Context Protocol (MCP) server for telemetry within chat systems using Weights & Biases Weave</strong>
</div>
Overview
MCP Telemetry provides a simple interface for logging and tracking conversations between users and LLMs. It leverages the Model Context Protocol to expose telemetry tools that can be used to trace and analyze conversations.
Features
- Start tracing sessions with custom identifiers
- Log comprehensive conversation data including:
- User inputs
- LLM responses
- LLM actions
- Tool calls and their results
- Seamless integration with Weights & Biases Weave for visualization and analysis
- Real-time monitoring of conversation flows
- Export and share conversation analytics
Installation
First, get a WandB API Key from: https://wandb.ai/settings#api
This server can be installed by adding the following json to your Claude desktop config:
{
"mcpServers": {
"MCP Telemetry": {
"command": "uv", -- this needs to be the location where uv is available, check via 'which uv'
"args": [
"run",
"--with",
"mcp[cli]",
"--with",
"weave",
"mcp",
"run",
"~/mcp-telemetry/server.py"
],
"env": {
"WANDB_API_KEY": "..." -- get one from wandb.com
}
}
}
}
Usage
Once installed, the MCP Telemetry server will automatically start when you launch Claude. It will begin collecting telemetry data for all conversations. You can view your telemetry data in the Weights & Biases dashboard.
Basic Usage
1. Start a conversation with Claude
2. The server will automatically track:
- User messages
- LLM responses
- Tool calls and their results
- Conversation metadata
Configuration
The server can be configured through environment variables:
- WANDB_API_KEY - Your Weights & Biases API key (required)
Examples
Starting a Tracing Session
Prompt Claude to trace that conversation. Example: Log this conversation with MCP Telemetry, topic will be Cats
Viewing Telemetry Data
1. Log in to your Weights & Biases account
2. Navigate to your project
3. You'll see various visualizations including:
- Conversation flows
- Tool usage patterns
- Response times
- Error rates
Contributing
Contributions are welcome! Please feel free to submit a Pull Request.
License
This project is licensed under the MIT License - see the LICENSE file for details.
Sign in to leave a review
Use Google, GitHub, or an email account so ratings stay tied to real people.
No reviews posted yet.
