No user-submitted related servers found.
Scored across 5 tools
Each tool has a documented role: media_probe inspects metadata, media_overview builds the grid/transcript, media_zoom reads one frame, media_locate searches text for a timestamp/page, and video_subtitles dumps subtitle segments. The overlap is mild but real: video_subtitles and media_overview both return timestamped text, and the descriptions spend considerable effort telling the agent which one to prefer. That overlap is resolvable from the docs, so boundaries are mostly clear.
Four of five tools use a clean media_<verb> pattern (media_probe, media_overview, media_zoom, media_locate). video_subtitles breaks the prefix and is also slightly misleading since it works on any media container, not just video. Otherwise consistent snake_case with predictable verbs.
Five tools map cleanly onto a probe -> overview -> zoom/search/transcript inspection pipeline, and each earns its place against a distinct cost model (free metadata, cheap grids, expensive single frames). Nothing is redundant or missing at the count level; it is well within the 3-15 sweet spot.
The lifecycle is covered: inspect, overview, drill down, search, and read subtitles, with ASR/OCR availability surfaced by media_probe and consumed by the other tools. Minor gaps remain: there is no explicit OCR/document-text tool despite OCR being reported, and no way to discover or list local media files, so the agent must already know absolute paths. These are workable around but not fully closed.