Help us improve
Share bugs, ideas, or general feedback.
From twelvelabs
Searches indexed videos using a reference image, optionally combined with text to refine results. Useful when users provide a picture or screenshot and want to find matching video content.
npx claudepluginhub twelvelabs-io/twelve-labs-claude-code-plugin --plugin twelvelabsHow this skill is triggered — by the user, by Claude, or both
Slash command
/twelvelabs:image-searchThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Search indexed videos using an image combined with optional text for more precise results. The text narrows image-based results — for example, an image of a car plus "red color" finds only red versions of that car model.
Searches indexed videos using natural language queries via TwelveLabs MCP. Extracts search terms and displays matching segments with timestamps.
Ingests video/audio from files, URLs, RTSP feeds, or desktop capture; indexes visual/spoken content for search; transcodes, edits timelines, generates assets, and creates real-time alerts.
Ingests video/audio from files, URLs, RTSP feeds, or desktop capture; builds visual/spoken/temporal indexes; searches moments with timestamps; edits timelines with subtitles, overlays, and audio; transcode/normalize and generate media assets.
Share bugs, ideas, or general feedback.
Search indexed videos using an image combined with optional text for more precise results. The text narrows image-based results — for example, an image of a car plus "red color" finds only red versions of that car model.
Use this skill when the user:
From the user's request, extract:
Image-only search (find content similar to the image):
With a URL:
Tool: mcp__twelvelabs-mcp__search
Parameters:
queryMediaUrl: "<image URL>"
queryMediaType: "image"
With a local file:
Tool: mcp__twelvelabs-mcp__search
Parameters:
queryMediaFile: "<absolute path to image file>"
queryMediaType: "image"
Composed search (image + text refinement):
With a URL:
Tool: mcp__twelvelabs-mcp__search
Parameters:
query: "<text description>"
queryMediaUrl: "<image URL>"
queryMediaType: "image"
With a local file:
Tool: mcp__twelvelabs-mcp__search
Parameters:
query: "<text description>"
queryMediaFile: "<absolute path to image file>"
queryMediaType: "image"
To search a specific index, add indexId.
Format results the same way as regular text search:
For each video with matching segments, display:
Example output:
Found matches for image search with "red color":
**demo_video.mp4**
Stream: https://stream.url/video.m3u8
Matching segments:
- 00:12 - 00:28 (16 seconds)
- 01:45 - 02:03 (18 seconds)
Same as regular search:
MM:SSHH:MM:SS| Image | Text | Finds |
|---|---|---|
| Photo of a car | "red color" | Red versions of that car model |
| Photo of a building | "at night" | That building during nighttime |
| Product photo | "premium edition" | Premium version of that product |
| Person's outfit | "on stage" | Similar outfits in stage scenes |
User: "Find scenes in my videos that look like this image: https://example.com/sunset.jpg" Action: Image-only search with queryMediaUrl
User: "Search for this car but in blue: https://example.com/car.jpg" Action: Composed search with query "blue color" + queryMediaUrl
User: "Find this building at night: https://example.com/building.jpg" Action: Composed search with query "at night" + queryMediaUrl
queryMediaUrl) and local file paths (queryMediaFile)visual search option