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

The description discloses significant behavioral traits beyond the annotations: staging-only semantics (writes to AMZ Vault, never hits Amazon), input normalization ('any casing is accepted and normalized to the correct per-API wire form'), level constraints (SB/SD require ad_group_id), and ad-type restrictions (SB is keyword-only, SD is product/brand-only, SB has no broad). These behaviors align with readOnlyHint=false and destructiveHint=false — the tool writes a staging record but is non-destructive — so there is no contradiction.

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 every sentence earns its place — item formats, level rules, match_type normalization, and staging semantics are all load-bearing. It front-loads the purpose before the constraints and caps off with the critical 'NOTHING is sent to Amazon' warning. The single-paragraph wall of text could be more scannable with structured separation, but the complexity of the three-ad-type matrix justifies the length.

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 tool with substantial domain complexity (SP/SB/SD distinctions, level requirements, match_type variants, staging semantics), the description covers the essential calling contract completely. Minor gaps exist: it doesn't reference how to review pending staged items (sibling staged_changes_pending), and doesn't mention item limits or duplicate handling. With no output schema, a bit more detail on the reviewable-diff result would round it out, but nothing critical 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?

With schema coverage at 67%, the description carries the semantic load and does so thoroughly. It explains exactly which fields pair with which ad type (campaign_id + keyword_text for SP/SB, asin_target for SP/SD, brand_id for SD), which level each ad type requires, and the full match_type value set with its default and normalization rules. This goes well beyond the raw schema, giving the agent the combinatorial knowledge needed to construct valid items.

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+resource pair — 'Propose adding negative keywords / negative ASIN targets' — which precisely states the operation and its object. It also distinguishes itself from the sibling stage_negatives_remove by framing this as the 'add' counterpart, and the 'STAGING ONLY' framing separates it from direct-mutation tools like confirm_staged_changes.

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 clearly establishes the staging workflow: changes are saved as a reviewable diff and 'NOTHING is sent to Amazon until confirm_staged_changes is called,' which tells the agent when this tool is appropriate (proposing changes without committing) versus when direct modification or confirmation is needed. It does not explicitly name alternatives like stage_negatives_remove or staged_changes_pending, but the add-vs-remove and stage-vs-confirm contrasts are strongly implied.

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