---
name: videograb-jobs
description: "Step-by-step job recipes for the Video Grab MCP server: Search freely licensed video; Export your own videos; Video metadata and captions. Use when the user asks things like \"Find a freely licensed video of ocean waves I can reuse, and give me the attribution.\" or \"How do I download my own YouTube videos from YouTube Studio?\""
license: MIT
---

# Video Grab jobs

Each section below walks through one job with the tools of the `videograb` MCP server. The same recipes are MCP prompts (prompts/list) on `https://api.openkrill.app/gateway/mcp`.
Connect the server at `https://videograb.openkrill.app/mcp` in your MCP client settings (remote streamable HTTP, no key), then call its tools as normal tool calls. They are not command line programs: Bash, curl and scripts cannot reach them.

## Search freely licensed video

Job: Search freely licensed video, with the Video Grab MCP server.

Use the `search_free_videos` tool for this job.

Work on the user's request. If there is none yet, ask what they need. A request this job handles: "Find a freely licensed video of ocean waves I can reuse, and give me the attribution."

Steps:
1. Call `search_free_videos` as a normal MCP tool call, with the request's values in the fields of its inputSchema. It needs `query`.
2. If a value the tool needs is missing and you cannot tell it from the request, ask the user one short question first.
3. This input is known to work; copy its format, not its values: `{"query":"ocean waves"}`
4. For the request "Find a freely licensed video of ocean waves I can reuse, and give me the attribution.", a correct answer looks like this: status ok, results with a download link, a free licence, attribution text, the source and an as_of date.
5. Follow the server's own rules:
   - Use search_free_videos for reusable footage on a topic.
   - It returns download links only for freely licensed files on Wikimedia Commons and the Internet Archive, each with a licence and attribution.
   - Pexels and Pixabay are searched only when their API keys are configured; when notes say they were not searched, do not invent results from them.
   - For any other platform, do not call it: tell the user to use that platform's own account settings.
   - Give the user download_url; files lists every direct file and captions lists caption files.
   - A YouTube answer is title and channel only, with no download and no captions.
   - We do not rip videos.
   - Downloading from TikTok, YouTube, Instagram, X, Douyin, Bilibili or Xiaohongshu, and removing watermarks, breaks those sites' terms.
   - Use the platform's own export for a video you uploaded, or search for a freely licensed file.
   - When a tool returns status not_offered, say that explanation and do not look for another downloader.
   - When status is not ok, follow error.hint: it names the next tool or input. not_free means give no download link and offer search_free_videos instead.
   - A missing caption file means the source did not list one, or the API does not allow the file.
   - The licence is what the source record states, so tell the user to check the item page before reuse.
   - Do not invent a download link, a watermark-free file, or a claim that a platform allows third-party downloading.
6. If the call fails or the answer looks wrong, tell the user, then report it with `submit_feedback` (kind bug, tool `search_free_videos`).

## Export your own videos

Job: Export your own videos, with the Video Grab MCP server.

Use the `own_video_export` tool for this job.

Work on the user's request. If there is none yet, ask what they need. A request this job handles: "How do I download my own YouTube videos from YouTube Studio?"

Steps:
1. Call `own_video_export` as a normal MCP tool call, with the request's values in the fields of its inputSchema. It needs `platform`.
2. If a value the tool needs is missing and you cannot tell it from the request, ask the user one short question first.
3. This input is known to work; copy its format, not its values: `{"platform":"youtube"}`
4. For the request "How do I download my own YouTube videos from YouTube Studio?", a correct answer looks like this: status ok, the YouTube Studio steps, the help URL https://support.google.com/youtube/answer/56100, and a line that other people's videos are not downloaded.
5. Follow the server's own rules:
   - It returns download links only for freely licensed files on Wikimedia Commons and the Internet Archive, each with a licence and attribution.
   - Pexels and Pixabay are searched only when their API keys are configured; when notes say they were not searched, do not invent results from them.
   - Use own_video_export when the user wants to download videos they uploaded to TikTok, YouTube, Instagram or X, and quote the steps and the help_url.
   - For any other platform, do not call it: tell the user to use that platform's own account settings.
   - Give the user download_url; files lists every direct file and captions lists caption files.
   - A YouTube answer is title and channel only, with no download and no captions.
   - We do not rip videos.
   - Downloading from TikTok, YouTube, Instagram, X, Douyin, Bilibili or Xiaohongshu, and removing watermarks, breaks those sites' terms.
   - Use the platform's own export for a video you uploaded, or search for a freely licensed file.
   - When a tool returns status not_offered, say that explanation and do not look for another downloader.
   - A missing caption file means the source did not list one, or the API does not allow the file.
   - The licence is what the source record states, so tell the user to check the item page before reuse.
   - Do not invent a download link, a watermark-free file, or a claim that a platform allows third-party downloading.
6. If the call fails or the answer looks wrong, tell the user, then report it with `submit_feedback` (kind bug, tool `own_video_export`).

## Video metadata and captions

Job: Video metadata and captions, with the Video Grab MCP server.

Use the `video_metadata` tool for this job.

Work on the user's request. If there is none yet, ask what they need. A request this job handles: "What licence and attribution should I use for the Commons file File:Ocean_surface_waves_06.ogv?"

Steps:
1. Call `video_metadata` as a normal MCP tool call, with the request's values in the fields of its inputSchema. It needs `ref`.
2. If a value the tool needs is missing and you cannot tell it from the request, ask the user one short question first.
3. This input is known to work; copy its format, not its values: `{"ref":"https://commons.wikimedia.org/wiki/File:Big_Buck_Bunny_4K.webm"}`
4. For the request "What licence and attribution should I use for the Commons file File:Ocean_surface_waves_06.ogv?", a correct answer looks like this: status ok, the Commons licence, attribution, a direct file URL, the source and an as_of date.
5. Follow the server's own rules:
   - It returns download links only for freely licensed files on Wikimedia Commons and the Internet Archive, each with a licence and attribution.
   - Pexels and Pixabay are searched only when their API keys are configured; when notes say they were not searched, do not invent results from them.
   - For any other platform, do not call it: tell the user to use that platform's own account settings.
   - Use video_metadata for one known item: a Commons file, an Internet Archive item, or a YouTube watch URL.
   - Give the user download_url; files lists every direct file and captions lists caption files.
   - A YouTube answer is title and channel only, with no download and no captions.
   - We do not rip videos.
   - Downloading from TikTok, YouTube, Instagram, X, Douyin, Bilibili or Xiaohongshu, and removing watermarks, breaks those sites' terms.
   - Use the platform's own export for a video you uploaded, or search for a freely licensed file.
   - When a tool returns status not_offered, say that explanation and do not look for another downloader.
   - A missing caption file means the source did not list one, or the API does not allow the file.
   - The licence is what the source record states, so tell the user to check the item page before reuse.
   - Do not invent a download link, a watermark-free file, or a claim that a platform allows third-party downloading.
6. If the call fails or the answer looks wrong, tell the user, then report it with `submit_feedback` (kind bug, tool `video_metadata`).