Distributed Video QA MCP

Search inside videos

video_search

Read-only

Find relevant video moments by searching spoken transcripts, keywords, and visual frames together, returning ranked timestamped hits for each match.

Instructions

Ranked hits from three channels fused by weighted reciprocal rank: meaning-based transcript search, keyword (BM25) search, and text-to-image search over frames. 'The slide with the pricing table' finds the frame even if nobody said it. Each hit lists the channels that matched it, with rank and raw score.

Tool call template

Replace each placeholder with a value matching the input schema. Only required parameters are included; add optional parameters as needed.

{
  "id": 1,
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "arguments": {
      "query": "<query>"
    },
    "name": "video_search"
  }
}

Input Schema

Name	Required	Description
`end`	No	Only before this second.
`query`	Yes	What to look for, in natural language.
`start`	No	Only after this second.
`top_k`	No	Hits to return.
`video_ids`	No	Limit to these video ids. Omit for every indexed video.
`modalities`	No	Search what was said, what was shown, or both.

Output Schema

Name	Required	Description
`hits`	Yes	Best first.
`count`	Yes	Hits returned.
`query`	Yes	The query, or the image path for image search.

Schema Changelog

Changes observed during successful MCP inspections.

First observedv0.3.0Nov 13, 2026

TDQS

A3.9/5.0

Behavior4/5

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

Annotations already cover the safety profile (readOnlyHint=true, openWorldHint=false), and the description adds genuine behavioral detail: results are fused by weighted reciprocal rank and each hit reports which channels matched plus rank and raw score. It omits practical traits like latency/cost of the frame channel or behavior when a video lacks indexed frames.

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

Conciseness4/5

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

Three tight sentences with zero filler, and the capability statement leads. It does open with mechanism ('weighted reciprocal rank') before stating plainly what the tool returns, which is slightly jargon-first for an agent skimming for a verb.

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

Completeness4/5

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

With a full output schema, complete parameter descriptions, and safety annotations, the description only needs to explain the retrieval model and hit shape, which it does. Minor gaps remain around prerequisites (must frames/transcripts be indexed) and cost of the visual channel.

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

Parameters3/5

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

Schema description coverage is 100% (query, start, end, top_k, video_ids, modalities all documented), so the schema carries parameter meaning and baseline is 3. The description's mention of 'channels' and 'rank and raw score' loosely maps to modalities and top_k but adds no new syntax or format guidance.

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?

The description concretely defines the operation as a fused hit list from three retrieval channels (transcript semantics, BM25 keyword, text-to-image over frames), which is far more specific than the title 'Search inside videos'. It does not name a sibling such as video_search_by_image or video_find_phrase, so the boundary against single-channel tools is left implicit.

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

Usage Guidelines4/5

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

The 'slide with the pricing table' example clearly signals the case this tool exists for: queries whose answer may be visual rather than spoken, so the agent knows to prefer it over pure transcript search. There is no explicit when-not-to-use statement or named alternative for image-only or exact-phrase lookups.

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