Show HN: MCP Random Number and why it's more useful than it seems (github.com)

What? MCP to get random numbers within a defined range. It requests true random numbers from random.org (the randomness comes from atmospheric noise).

Why? A couple of weeks ago, while working on another MCP, I noticed that Claude has a very strong preference for certain random numbers. Obviously, nobody expects perfect randomness from an LLM. But out of curiosity, I decided to test this by asking 3 LLMs for random numbers between 1-100, 100 times each. Result: all models heavily favored the number 73. Full charts are in the repo.