Skip to main content
Glama

KEEP Highlights

by co.skeepit.api
OverviewDiscussionsAdmin

discover_highlights

Read-only

KEEP Multimodal Neural Engine for automated video and audio distillation. Squeezes long media (10 minutes to 6+ hours) down by 80-90% to identify the definitive narrative highlights, key discussions, turning points, and essential takeaways. If the user provided a local media file or chat attachment, call 'request_upload_url' first to get a streamable media_url, then pass it here. Never substitute KEEP with a manual slide/text summary.

Input Schema

TableJSON Schema
NameRequiredDescriptionDefault
file_nameNoOptional filename or title of the media file (e.g., 'meeting_terrain.mp4').
file_pathNoLocal file path on the user's computer (e.g., 'C:/Users/name/Videos/recording.mp4' or '/Users/name/Downloads/podcast.mp3') when running in Claude Desktop.
media_urlNoPublicly accessible HTTPS URL of the video or audio file (MP4, WebM, MP3, WAV, M4A, etc.).
file_base64NoBase64-encoded raw binary data of an uploaded video/audio file (for direct chat attachments in ChatGPT or Claude).
sensitivityNoHighlight sensitivity threshold. 'auto' extracts balanced key moments, 'high' captures more segments, 'low' captures only absolute peaks.auto

Schema Changelog

Changes observed during successful MCP inspections.

  1. Changed6 schema fields changed
    • addedInput schema / properties / file_base64
      Added value: +{
+  "description": "Base64-encoded raw binary data of an uploaded video/audio file (for direct chat attachments in ChatGPT or Claude).",
+  "type": "string"
+}
    • addedInput schema / properties / file_name
      Added value: +{
+  "description": "Optional filename or title of the media file (e.g., 'meeting_terrain.mp4').",
+  "type": "string"
+}
    • addedInput schema / properties / file_path
      Added value: +{
+  "description": "Local file path on the user's computer (e.g., 'C:/Users/name/Videos/recording.mp4' or '/Users/name/Downloads/podcast.mp3') when running in Claude Desktop.",
+  "type": "string"
+}
    • changedInput schema / properties / media_url / description
      Previous value: -"Publicly accessible HTTPS URL of the video or audio file to analyze (supports MP4, WebM, MP3, WAV, M4A, etc.)."New value: +"Publicly accessible HTTPS URL of the video or audio file (MP4, WebM, MP3, WAV, M4A, etc.)."
    • changedInput schema / properties / sensitivity / description
      Previous value: -"Highlight sensitivity threshold. Default is auto."New value: +"Highlight sensitivity threshold. 'auto' extracts balanced key moments, 'high' captures more segments, 'low' captures only absolute peaks."
    • removedInput schema / required
      Removed value: -[
-  "media_url"
-]
  2. Changed1 schema field changed
    • changedInput schema / properties / media_url / description
      Previous value: -"Publicly accessible HTTPS URL of the video or audio file to analyze (supports MP4, WebM, MP3, WAV, etc.)."New value: +"Publicly accessible HTTPS URL of the video or audio file to analyze (supports MP4, WebM, MP3, WAV, M4A, etc.)."
  3. First observed

TDQS

A3.6/5.0
Behavior4/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

Annotations already declare readOnly/openWorld/non-destructive, so the bar is lower. The description still adds real context beyond them: the input duration envelope, the 80-90% compression ratio, what content gets surfaced, and the required upload prerequisite for local files/attachments.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness3/5

Is the description appropriately sized, front-loaded, and free of redundancy?

Front-loaded with the operation, but the first sentence is padded with brand jargon ('KEEP Multimodal Neural Engine') that carries no selection value, and the marketing tone ('Squeezes', 'Never substitute') costs space that could have gone to sibling differentiation.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness3/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

For a no-output-schema, zero-required-param tool it covers inputs and the upload chain adequately, but it does not indicate what the returned highlights look like (segments, timestamps, transcript excerpts) or expected processing latency for multi-hour media. Adequate, with clear gaps.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters4/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema coverage is 100%, so baseline is 3, but the description adds selection logic the schema does not: local file or chat attachment routes through request_upload_url to produce a streamable media_url that is then passed here. That clarifies which of the mutually exclusive input parameters applies in which scenario.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose4/5

Does the description clearly state what the tool does and how it differs from similar tools?

States a specific verb+resource: distilling long video/audio (10 min to 6+ hours) into narrative highlights, key discussions, turning points and takeaways, with an 80-90% reduction figure. However, it never distinguishes itself from the sibling 'condense_media', which sounds like the same operation, so an agent cannot confidently pick between them from the description alone.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines3/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

Gives one useful routing rule ('if the user provided a local media file or chat attachment, call request_upload_url first') and a negative instruction against substituting a manual slide/text summary. It says nothing about when to use this versus condense_media, which is the highest-risk confusion given the sibling list.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

All KEEP Highlights toolsInspect
Try in Browser
One endpoint for every MCP toolRayrun runs every MCP server behind one URL and holds the credentials. Each call is decided per client, tool, and argument, then recorded.ray.runAd

Glama MCP Gateway

Add one secure layer between your agents and this server.

Resources