mcp-telemetry
MCP Telemetry is a server for logging and tracking conversations in chat systems, utilizing the Model Context Protocol. It offers tools for real-time monitoring and analytics, with support for visualization through Weights & Biases Weave.
Overview
MCP Telemetry is a Model Context Protocol (MCP) server designed for telemetry within chat systems. It provides tools to log and track conversations between users and large language models (LLMs) using Weights & Biases Weave. The features include tracing sessions with custom identifiers, logging user inputs, LLM responses, actions, tool calls, and results, along with integration for visualization, real-time monitoring, and analytics exportation.
Features
- Custom identifier tracing sessions
- Comprehensive data logging including inputs, responses, actions, and tool calls
- Integration with Weights & Biases Weave
- Real-time monitoring and analytics export
Installation
- Obtain a WandB API Key.
- Install the server by modifying the Claude desktop configuration.
Usage
- The MCP Telemetry server starts automatically with Claude, collecting telemetry data.
- View telemetry data on the Weights & Biases dashboard.
Configuration
Configure using environment variables, notably the WANDB_API_KEY for Weights & Biases.
Contributing
Contributions are welcome via Pull Requests.
License
Licensed under the MIT License.