search_timestamp

Describe what you’re looking for in plain language, and the tool returns the timestamp where that event, object, or phrase appears in the video.

Parameters

Parameter	Type	Required	Description
videoPath	string	✓	Path to the video, file:// URI, or remote URL
query	string	✓	Natural language description of what to find
provider	string	—	Override the AI provider: gemini, m3, kimi, qwen, mimo, glm

Usage Examples

Find a gesture:

“At what timestamp does the person wave in ./clip.mp4?”

Find spoken content:

“At what point does the speaker mention pricing in https://www.youtube.com/watch?v=abc123?”

Find a visual element:

“When does the code editor first appear in ./tutorial.mp4?”

Find a scene transition:

“At what timestamp does the presentation switch from slides to a live demo in ./webinar.mp4?”

Find a person:

“When does Alice appear on screen in ./meeting.mp4?”

How It Works

• Gemini uploads the full video and asks the model directly for the exact timestamp(s).
• All other providers extract frames at a fixed interval (scaled to the video length, 2–15 s), ask the model which frame index best matches the query, and convert the index back to a timestamp.

Return Value

The tool returns:

• Best match timestamp in seconds (e.g. near 47.5s)
• Estimated range around the match (± half the sampling interval)
• The model’s response explaining the match
• The matched frame as an inline image

If nothing matching is found, the model states that clearly in its response.