Fast Video Cataloger
OfficialFast Video Cataloger MCP server
Search, view and tag your local video library from an AI assistant.
Fast Video Cataloger keeps a searchable catalog of your own video files on Windows: scene thumbnails, keywords, transcripts and faces. This MCP server lets an assistant such as Claude work with that catalog. You ask for footage in your own words; the assistant searches, looks at the actual shots and does the filing for you.
Find videos and moments by what was said, with timecodes
Find scenes by describing the picture ("a drone shot over a city at night"), nothing tagged first
Fetch scene thumbnails so the assistant can see and judge a shot
With an Editor key: add or remove keywords, build bins, index new videos
It runs against the Fast Video Cataloger server on your own computer. No video is uploaded; the assistant receives the titles, transcript lines and scene images it asks for.
More: MCP server for your video library
Requirements
Windows 10 or 11 (x64)
Fast Video Cataloger 10.4 or later. The free 30-day trial works.
fvc-mcp.exeis also installed with the program, together with the .NET 10 runtime it needs.Your catalog shared through the Fast Video Cataloger server: in the program, open the start page's Server section and click Setup. The assistant connects to the server, not to the desktop program, so it works with the program closed.
An API key: Manage Users in the same Server section.
Viewerto search and read,Editorto also tag and build bins.
Related MCP server: semantic-image-search-mcp
Install
Claude Desktop: one click
Download fvc-mcp-<version>.mcpb from Releases and
open it. Claude Desktop asks for the server address (leave http://localhost:8754 when the server
runs on this computer) and your API key.
Claude Desktop: config file
Use the copy installed with Fast Video Cataloger. In claude_desktop_config.json:
{
"mcpServers": {
"fast-video-cataloger": {
"command": "C:\\Program Files\\FastVideoCataloger\\fvc-mcp.exe",
"env": {
"FVC_SERVER_URL": "http://localhost:8754",
"FVC_API_KEY": "fvc_your_key_here"
}
}
}
}Claude Code
claude mcp add fast-video-cataloger --env FVC_SERVER_URL=http://localhost:8754 --env FVC_API_KEY=fvc_your_key_here "--" "C:\Program Files\FastVideoCataloger\fvc-mcp.exe"Keep the quotes around "--" in PowerShell.
Gemini CLI
gemini mcp add fast-video-cataloger "C:\Program Files\FastVideoCataloger\fvc-mcp.exe" -s user -e FVC_SERVER_URL=http://localhost:8754 -e FVC_API_KEY=fvc_your_key_hereFull setup, things to try and troubleshooting: Connect an AI Assistant (MCP).
Configuration
Variable | Meaning |
| Address of your catalog server. Default |
| The API key. |
|
|
Tools (23)
Tool | What it does |
| Find videos by free text or by keyword. |
| Find individual scenes, with their timecodes. |
| Find scenes by describing the picture, or scenes that look like a given one. Needs the visual search index and, for descriptions, the Visual search model on the server. |
| Find spoken words, with their video and timecode. |
| Find people by name or tag. |
| Every keyword in the catalog, so the assistant knows what it can search for. |
| One catalog entry in full. |
| A video's scene thumbnails in timecode order. |
| The transcript as timed lines. |
| Keywords and cast. |
| The thumbnail image itself, so the assistant can see the shot. |
| Catalog size, the build serving it, and existing bins. |
| Add keywords to a video or a single scene. Editor key. |
| Remove named keywords again. Editor key. |
| Build a collection. Editor key. |
| Index catalog entries that have no thumbnails yet. Editor key. |
It cannot delete videos or bins, take a video out of a bin, play video or drive the player.
Source
src/FvcMcpServer is the source of the fvc-mcp.exe shipped with Fast Video Cataloger, copied
from each release. It is a thin client over the Fast Video Cataloger server's REST API: it holds
no state and every tool is one or two REST calls. Build with the .NET 10 SDK:
dotnet publish src/FvcMcpServer -c Release -r win-x64 --self-contained false -p:PublishSingleFile=trueReleased builds are code-signed by VideoStorm Sweden AB. Issues and pull requests are welcome; the code here follows the version in Fast Video Cataloger, so accepted changes ship with the next release.
License
MIT for the source in this repository, see LICENSE. Bundled packages: THIRD-PARTY-NOTICES.md. Fast Video Cataloger itself is commercial software with a free 30-day trial.
Available Tools
23 toolsadd_video_to_binAdd a video to a binA
Put a video into a bin. Needs an API key with the Editor role.
Inputs: binId (required), videoId (required).
catalog_statsCatalog sizeARead-only
How many videos, actors and images the Fast Video Cataloger catalog holds. This is the user's local video library, not cloud storage or the file system. Useful as a first call to confirm the connection works and to judge how broad a search should be.
Inputs: no named parameters.
create_binCreate a binA
Create a new bin to collect videos in. Returns the new bin including its id. Needs an API key with the Editor role.
Inputs: label (required), parentId.
get_bin_videosGet a bin's videosARead-only
List the videos collected in one bin. A bin can hold the whole catalog, so this returns the first 50 by default - check totalCount and page with offset.
Inputs: binId (required), limit, offset.
get_scene_imageLook at a sceneARead-only
Fetch the actual image of a scene thumbnail so you can see what is in the shot. Get thumbnail ids from search_scenes, search_scenes_semantic or get_video_scenes.
Inputs: thumbnailId (required).
get_videoGet a videoBRead-only
Get one catalog entry by its id, including title, description, path, length and rating.
Inputs: videoId (required).
get_video_actorsGet a video's castCRead-only
List the people recorded as appearing in a video.
Inputs: videoId (required).
get_video_scenesGet a video's scenesARead-only
List the scene thumbnails of one video in timecode order. Use this to see how a video is structured before deciding which moment to look at. Returns totalCount so you can tell whether there are more than you asked for.
Inputs: videoId (required), limit, offset.
get_video_tagsGet a video's tagsBRead-only
List the keywords a video is tagged with.
Inputs: videoId (required).
get_video_transcriptGet a transcriptARead-only
Get the transcript of a video as timed lines. Empty if the video has not been transcribed. A feature-length transcript is very long, so this returns the first 200 lines by default - check totalCount and page with offset if you need the rest.
Inputs: videoId (required), limit, offset.
index_videoIndex a videoA
Ask the server to index a video that is in the catalog but has no scene thumbnails yet. Extracts thumbnails and duration, plus whatever else the server is configured for such as transcription. This takes minutes and runs in the background - poll get_video_scenes to follow it. Needs the Editor role.
Inputs: videoId (required).
list_binsList binsARead-only
List the bins in the user's Fast Video Cataloger catalog. A bin is a user-made collection of videos, like a folder of shortlisted clips.
Inputs: no named parameters.
list_keywordsList the catalog's keywordsARead-only
List every keyword that exists in the whole Fast Video Cataloger catalog, rather than the keywords on one video. Use it to find out what you can actually search for before guessing - a keyword that is not in the catalog simply returns nothing. This is the vocabulary only, with no counts, and it mixes keywords the user typed with ones written automatically by object detection, so a catalog can list things like 'zebra' or 'toaster' that appear nowhere in the footage. Both scopes can contain either kind. To find out whether a keyword is really used, search for it.
Inputs: scope.
list_videos_needing_indexingList unindexed videosARead-only
List catalog entries that have no scene thumbnails or no known length yet. Pass each id to index_video to work through them. A newly added folder can leave thousands pending, so this returns the first 50 by default - check totalCount and page with offset to walk the rest.
Inputs: limit, offset.
search_actorsSearch actorsARead-only
Search the people recorded in the user's Fast Video Cataloger catalog, by name or by the tags on them. These are the cast of their own footage, not contacts.
Inputs: tags, limit, query.
search_scenesSearch scenesARead-only
Search individual scenes inside the user's Fast Video Cataloger videos, rather than whole videos. This is how you find the moment inside a video: each result carries a thumbnail id, the video it belongs to, and its timecode. Pass the thumbnail id to get_scene_image to look at it.
Inputs: limit, query, videoId, keywords.
search_scenes_semanticFind scenes by descriptionARead-only
Find scenes by describing what is in the picture - 'a dog running on a beach', 'close-up of hands typing', 'city skyline at night'. Nothing has to be tagged: every indexed scene is ranked by how well it matches the words, so use this when the keywords search_scenes needs do not cover what you are after, and use similarTo to find footage that looks like a scene you already have (the source scene itself is left out of those results). Results come best first, each with the thumbnail id, the video, the time (seconds and hh:mm:ss) and a relevance score. Judge by the order, not the number: for a description a good match scores only about 0.15-0.30, while scenes that look like a given one score 0.8 and above, and scores are comparable only within one search. Pass a thumbnail id to get_scene_image to check what a scene really shows. If the answer says the model or the index is missing, tell the user what to do; do not retry.
Inputs: limit, query, videoId, similarTo.
search_transcriptsSearch spoken wordsARead-only
Search what is said out loud in the user's Fast Video Cataloger videos. Returns the matching transcript lines themselves - each with the id of the video it belongs to (videoFileID) and its start and end time in seconds, so you can go straight to the moment. Use get_video_transcript to read the lines around a hit. Only videos that have been transcribed are searched.
Inputs: query (required), limit.
search_videosSearch videosARead-only
Search the user's Fast Video Cataloger catalog - their local library of video files, not cloud storage or the file system. Use this for any question about their videos, footage or clips. Use 'query' for a free text search over titles, descriptions and keywords, or 'keywords' for a comma-separated keyword match. Returns catalog entries including each video's id, title, path and length.
Inputs: limit, query, actors, offset, +3 more.
tag_sceneTag a sceneA
Add keywords to a single scene thumbnail rather than the whole video. Needs an API key with the Editor role.
Inputs: tags (required), thumbnailId (required).
tag_videoTag a videoA
Add keywords to a video. Existing tags are kept. Needs an API key with the Editor role.
Inputs: tags (required), videoId (required).
untag_sceneRemove a scene's keywordsADestructive
Remove keywords from a single scene thumbnail - the counterpart to tag_scene. Only the named keywords are removed from that one scene; other scenes and the video's own tags are untouched. A keyword that no scene uses any more also disappears from the catalog's scene keyword list. Returns what was removed and what did not match. Needs an API key with the Editor role.
Inputs: tags (required), thumbnailId (required).
untag_videoRemove a video's keywordsADestructive
Remove keywords from a video - the counterpart to tag_video, for undoing a tag that was added in error. Name the keywords exactly as get_video_tags reports them; only those are removed and every other tag on the video is left alone. Other videos keep the keyword, but a keyword that no video uses any more also disappears from the catalog's keyword list. Returns what was removed and what did not match. Needs an API key with the Editor role.
Inputs: tags (required), videoId (required).
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 23 tools
Each tool targets a clearly distinct resource or action, with little overlap. The various search tools (search_videos, search_scenes, search_transcripts, search_scenes_semantic) are well differentiated by the entity they search and the method used. Tagging and untagging tools are split by scope (video vs scene), and bin tools are distinct from video and scene tools.
Almost all tools follow a consistent verb_noun snake_case pattern (e.g., get_video, search_scenes, tag_video, list_bins). The only notable deviation is catalog_stats, which uses a noun_noun form without a verb. This is a minor inconsistency in an otherwise predictable naming scheme.
With 23 tools, the set falls into the borderline-heavy range for an MCP server. While the domain (video catalog with bins, scenes, transcripts, actors, tags, and indexing) is broad, some tools could potentially be consolidated (e.g., add/remove tag pairs), and the count feels high relative to the missing lifecycle operations.
The toolset covers search, tagging, bins, indexing, and transcript/scene access well, but lacks basic update and delete operations for core entities: there is no update_video, delete_video, delete_bin, or remove_video_from_bin. These gaps mean the surface is not a complete CRUD/lifecycle set for the catalog.
Maintenance
Related MCP Connectors
Search photos and video moments by visual meaning, people, dates, and folders across local libraries, Apple Photos, and OneDrive. Manage collections, groups, favorites, people, and duplicate review. Synced metadata remains searchable while the desktop is offline; local and Apple Photos media access needs the owning device online with remote access enabled. OneDrive media can be accessed directly from the cloud. OAuth or Bearer API keys. Paid Cloud Access required; hosted visual search uses prepaid credits. Text-in-image search is semantic matching, not guaranteed exact OCR. Setup: https://organizewith.ai/en/guides/connect-ai-assistants
Search your video library with natural language and retrieve relevant moments through MCP.
VideoDB lets an AI agent watch, search and clip video.
Your YouTube library in Claude, ChatGPT and Cursor: transcripts, breakdowns, summaries, search.
Related MCP Servers
- FlicenseNot gradedqualityDmaintenanceBridges Claude and video content by extracting keyframes and transcribing audio, enabling Claude to analyze video files.-
- AlicenseAqualityCmaintenanceEnables natural language search of local photo archives using AI-powered semantic understanding, with integration into Claude Desktop via the Model Context Protocol.42MIT
- AlicenseNot gradedqualityBmaintenanceSearch within videos by dialogue, on-screen visuals, and recurring faces using natural language - connected with your favorite agent (like Claude, Codex, Hermes).59MIT
- AlicenseBqualityCmaintenanceProvides Claude Code with local ffmpeg-based tools to probe, trim, concat, transform, overlay, mix audio, render JSON timelines, and run automated QA on video files—no API key required.9Apache 2.0