Show HN: EchoChain to see what happens when LLMs play telephone

Category: ai-ml

Tags: llm, evaluation, cli, openrouter, semantic-drift

Score: 4.3/10 (Innovation: 5, Technical: 4, Documentation: 5, Utility: 3)

echoChain is a Python CLI tool that measures how an LLM's answer drifts when models rewrite it in a loop—essentially automated telephone game for evaluating semantic stability. It provides run/analyze commands, configurable cycles, and optional sentence-transformer similarity scoring, making it a niche but interesting experiment for LLM robustness research.

Target audience: AI researchers and ML engineers studying model robustness and self-consistency

Repository: https://github.com/smusali/echoChain · Python · MIT

View on Hacker News