watch_video

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

The description goes well beyond the annotations. It discloses that the tool is fully offline, uses platform subtitles when available otherwise local Whisper, caches results per video, has specific time expectations (5–60s short subtitled, minutes for long unsubtitled), downloads a model on first run (~480MB CPU / ~1.6GB GPU), and explains the RUNNING reply behavior. It also states the reply format (English overview with metadata, chapters, contact sheets, file paths, transcript with [mm:ss] timestamps). This is rich behavioral context that annotations alone (readOnlyHint=false, idempotentHint=true, destructiveHint=false) do not provide.

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

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

The description is dense but well-organized, front-loading the core purpose and then covering input types, offline behavior, caching, timing, the RUNNING protocol, reply format, and exclusions. Every sentence carries useful information. It loses a point for being somewhat long and for burying the 'Not for live streams or images' exclusion at the end, which is important routing information that could be more prominent.

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

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

For a complex tool with 10 parameters, no output schema, and no sibling differentiation in the schema, the description is remarkably complete. It covers input types, platform support, offline behavior, caching, timing, model download, the RUNNING retry protocol, reply format, timestamp citation format, and exclusions. It also references sibling tools (get_frames, get_transcript) for follow-up actions. Nothing an agent needs to call this tool correctly is missing.

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

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

Schema description coverage is 100%, so the baseline is 3. The description adds value beyond the schema by explaining the video parameter's accepted URL platforms (YouTube, Bilibili, Douyin, Xiaohongshu, TikTok, Vimeo) and clarifying that results are cached per video, which gives meaning to the refresh parameter. It also explains the max_chars truncation behavior and its relationship to get_transcript. However, it doesn't add much detail on mode, model, language, or output_dir beyond what the schema already documents, so a 4 rather than 5.

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

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

The description opens with a specific verb and resource: 'Turn a video into a timestamped transcript plus a keyframe overview' and immediately lists downstream uses (summarize, answer questions, extract steps/commands/code, write notes). It also names the input types (local file or specific platforms) and explicitly excludes live streams and images, which distinguishes it from sibling tools like get_frames or get_transcript.

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

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

The description gives explicit when-to-use context: use it to summarize, answer questions, extract steps/commands/code, or write notes. It also states when NOT to use it ('Not for live streams or images') and references sibling tools for follow-up actions ('see them with get_frames', 'start time for get_transcript'). It even provides operational guidance about the RUNNING reply and not starting other videos meanwhile.

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