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

Annotations already establish readOnlyHint=false, destructiveHint=true, and idempotentHint=false, so the write/destructive profile is known and the description adds genuinely new behavior: the legible-text guarantee, the artifact-only side-effect boundary (publishing delegated to send_to_user), the write-tier approval flow with its from-now-on vs just-once distinction, the do-not-re-ask rule, and the $0 cost. Minor gap: the description never clarifies what destructive effect justifies destructiveHint=true, and 'produces artifacts only' sits in mild tension with that flag.

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?

Purpose and quality guarantee are front-loaded, and each block carries distinct operational value. However, the legibility claim appears three times ('REAL, legible text', 'the text stays sharp', 'legible cited-answer overlay'), 'cited-answer overlay' twice, and the approval bracket is verbose ('the card is the yes — do not re-ask in chat'). Tightening the repetition would make it stronger.

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?

The workflow boundary (generate artifacts → hand off via send_to_user(intent:"publish")) and the approval/consent behavior are covered explicitly, which is the kind of operational context agents most often lack. However, with no output schema, the description never says what the call returns — artifact IDs, preview URLs, or references needed to feed send_to_user — leaving the agent to infer the handoff.

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 coverage is 100%, and the schema already contains unusually rich parameter guidance (citation authenticity rules, brand_label ownership, folder reuse, format dimensions, slide count guidance), so the baseline is 3. The description offers only light framing — 'structured slide copy' for the slides parameter and 'the cited-answer overlay' for citation — without adding any semantic depth beyond the schema.

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 first sentence names a specific deliverable ('multi-slide image carousel + a LinkedIn-PDF') built from 'structured slide copy', with a concrete differentiator (real, legible text rather than garbled AI-rendered text). The Routing line explicitly scopes the tool to 'Carousel / slide deck / LinkedIn PDF / legible cited-answer overlay', which lets an agent separate it from sibling tools like generate_video, generate_html_visual, and generate_vector_image without opening their schemas.

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 an explicit routing rule ('Carousel / slide deck / LinkedIn PDF / legible cited-answer overlay → use this'), names the intended use case ('value-demonstration B2B content'), and states precisely what the tool does not do ('Produces artifacts only; publish via send_to_user(intent:"publish")'). It also discloses the approval gating and tells the agent to call it 'on the first clear ask' and not re-ask. The only gap is the absence of an explicit named alternative for the negative case, though the artifact-only boundary mostly compensates.

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