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

Annotations already declare non-read-only, non-destructive, closed-world, so the description doesn't need to repeat the safety profile. It does usefully add the blocking-by-default behavior and the implicit fan-out when mediaId is omitted (a potentially long or costly operation). It omits the approval flow implied by approvalId and says nothing about cost, duration, or what happens to an existing transcript unless force is set (which only the schema documents).

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?

Three tight sentences, front-loaded with the verb and engine, then scope behavior, then the blocking default — no filler. The parenthetical feature list is dense but earns its place by signaling output capabilities that have no output schema backing them.

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 an 8-parameter, no-output-schema, non-destructive mutation that calls an external service, the description covers the two most consequential decisions (scope and blocking) but leaves diarization unexplained, the needs_approval/approvalId recovery path unmentioned, and the shape of the returned transcript undocumented. Adequate but with clear gaps for an agent that must call it correctly the first time.

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?

With 63% schema coverage, the description adds real meaning for two of the weakest documented parameters: mediaId's omission semantics (batch over all timeline media with audio) and wait's default blocking behavior. It does not touch diarization, which is undocumented in both schema and description, nor the approval prerequisite — keeping it below a 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 names a specific verb and resource ('Transcribe media') and adds concrete output characteristics — Soniox engine, word timestamps, speaker labels, 60+ languages. It does not explicitly distinguish itself from near-siblings such as generate_captions, voice_captions, or get_transcript, so the agent must infer the split.

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?

'Without mediaId transcribes every timeline media with audio' gives a useful selection rule for single-vs-batch scope, and 'wait=true (default) blocks until done' tells the agent about call posture. However, no alternative tool is named (generate_captions, voice_captions, get_transcript), and there is no when-not guidance such as avoiding batch mode on large timelines.

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