MCP Internet Speed Test
@inventer-dev
The MCP Internet Speed Test is a specialized Model Context Protocol (MCP) server that enables AI models and agents to perform internet speed measurements.
概览
What is MCP Internet Speed Test?
MCP Internet Speed Test is a Model Context Protocol (MCP) server that allows AI models and agents to measure, analyze, and report network performance metrics through a standardized interface. It uses the SpeedOf.Me incremental testing methodology with files from 128 KB to 128 MB and detects CDN providers such as Fastly, Cloudflare, and AWS CloudFront.
How to use MCP Internet Speed Test?
Install the package from PyPI with pip install mcp-internet-speed-test or use uvx mcp-internet-speed-test. Then run mcp-internet-speed-test to start the server. Configure it in MCP clients (e.g., Claude Desktop) by adding the command and optional arguments to the MCP configuration file.
Key features of MCP Internet Speed Test
- Smart incremental testing with configurable 8-second threshold
- Download and upload speed testing (128 KB to 128 MB)
- Latency and jitter measurement with configurable samples
- Multi-CDN detection (Fastly, Cloudflare, AWS CloudFront)
- Geographic location mapping (150+ POPs worldwide)
- Cache status analysis (HIT/MISS) and server header extraction
- Single
run_complete_testfunction for comprehensive metrics - Available via PyPI, uv, Docker, and local development
Use cases of MCP Internet Speed Test
- Network diagnostics to identify server proximity and routing issues
- Performance analysis correlating speed results with CDN locations
- CDN optimization by detecting suboptimal ISP routing
- Troubleshooting network problems such as firewall blocks or DNS misconfigurations
FAQ from MCP Internet Speed Test
What does MCP Internet Speed Test do?
It provides tools for AI models to measure download/upload speed, latency, jitter, and CDN server metadata via an MCP interface, using the SpeedOf.Me incremental testing methodology.
What are the dependencies and runtime requirements?
Requires Python 3.12 or higher and the packages mcp[cli]>=1.25.0 and httpx>=0.27.0. Internet access to test files (hosted on GitHub via Git LFS) and upload endpoints (Cloudflare Workers and HTTPBin) is needed.
Where do the test files and data live?
Download test files are stored in the GitHub repository inventer-dev/speed-test-files using Git LFS. Upload tests use Cloudflare Workers (httpi.dev) and HTTPBin (httpbin.org). No user data is stored; only real-time metrics are returned.
How is CDN server information obtained?
The server extracts headers like x-served-by, via, and x-cache from HTTP responses, then maps POP codes and locations using internal lookup tables for Fastly (50+ locations), Cloudflare, and AWS CloudFront.
Is there any authentication or transport security?
The README does not mention authentication or transport encryption beyond the normal HTTPS used by test endpoints. The MCP server itself can be run locally or in a container; no API keys or credentials are required.