zhaoyue722/llm-usage-mcp

Overview

A local-first, multi-provider tool written in Python that records LLM API usage and costs, then exposes that data to coding agents through the Model Context Protocol (MCP). It runs on your own machine to avoid sending spend data to third parties.

Best for

Best for
Developers using MCP-based coding agents who need local, multi-provider LLM cost tracking.

Use cases

• Track per-request or cumulative LLM API spend across multiple providers
• Surface real-time cost information inside MCP-compatible coding agents
• Monitor budget usage during development without switching tools

How to use

Install

uv tool install llm-usage-mcp

Tools exposed

• query_spend
• usage_summary
• compare_providers
• recommend_provider
• get_pricing
• list_providers
• record_usage
• LLM_USAGE_DB_URL
• LLM_USAGE_PROXY_PORT

Tested with

Claude Code, Cursor, ChatGPT

Notes

A local-first, multi-provider tool written in Python that records LLM API usage and costs, then exposes that data to coding agents through the Model Context Protocol (MCP). It runs on your own machine to avoid sending spend data to third parties.

3 stars on GitHub. Last updated 2026-07-13. Licensed MIT.

Use cases

• Track per-request or cumulative LLM API spend across multiple providers
• Surface real-time cost information inside MCP-compatible coding agents
• Monitor budget usage during development without switching tools

Pros

• Runs locally so spend data never leaves your environment
• Supports multiple LLM providers from a single integration
• Connects directly to MCP agents for in-flow visibility

Cons

• Requires manual setup and configuration for each provider
• Limited community adoption (3 GitHub stars) implies early-stage maturity
• Only exposes data through the MCP protocol, not a standalone dashboard

Indexed from awesome-mcp-servers-punkpeye and enriched against its public facts.