Show HN: Model Salt, fixing the pelican problem in LLMs by adding entropy

Category: ai-ml

Tags: mcp, llm, randomness, creativity, claude-code, codex

Score: 6.8/10 (Innovation: 7, Technical: 6, Documentation: 8, Utility: 6)

Model Salt is an MCP server that injects external randomness into LLM prompts by pre-seeding decisions (like the pelican's pose or a shop's name) from public entropy sources before the model generates output, so identical prompts yield genuinely varied results. It's clever because it turns a sampling problem into an explicit commitment step, and it ships with one-command setup for Claude Code, Codex, Gemini CLI, and Kimi Code.

Target audience: AI power users, prompt engineers, and developers building creative LLM workflows in MCP-compatible CLI clients

Repository: https://modelsalt.com/ · JavaScript · MIT · 1 stars

View on Hacker News