{"id":"kevinwatt-ffmpeg-mcp-lite","name":"ffmpeg-mcp-lite","af_score":59.8,"security_score":22.2,"reliability_score":26.2,"what_it_does":"ffmpeg-mcp-lite is an MCP server that exposes FFmpeg-based media tools to AI clients (e.g., Claude/Dive). It can get media metadata, convert formats, compress, trim, extract audio, extract frames, merge videos, and burn in subtitles.","best_when":"You need local/desktop integration of FFmpeg capabilities into an MCP-capable AI workflow and you can supply input files and manage FFmpeg/ffprobe binaries on your system.","avoid_when":"You need authenticated, network-accessible APIs with clear enterprise security controls, or you cannot install/run FFmpeg and an MCP server process.","last_evaluated":"2026-03-30T15:33:00.816526+00:00","has_mcp":true,"has_api":false,"auth_methods":["None explicitly documented (local MCP server invoked via client config)"],"has_free_tier":false,"known_gotchas":["Relies on local ffmpeg/ffprobe binaries (may fail if missing or misconfigured).","Media processing can be resource intensive (CPU/disk) and may be slow for large files.","Outputs may overwrite existing files depending on how output paths are handled (not documented).","Time parsing (timestamps vs seconds) is described but not formally specified; agents may mis-specify time formats."],"error_quality":0.0}