"author:rogermsc" matching MCP servers:

Turns silent defects in AI-generated video and audio — wrong speaking pace, mismatched loudness, truncation, dead air, late captions, broken layout — into hard assertion failures inside tests and CI. Enables agents to check individual files or whole output directories against platform presets, returning pass/fail exit codes and human-readable bug reports instead of raw measurements.

Provides an MCP server that gives LLM agents an evolutionary memory: entries pay upkeep and survive only if they earn their keep through measured outcomes, with tools for querying, settling, and managing memory.