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

Annotations already declare readOnlyHint=true and destructiveHint=false, yet the description adds substantial behavior beyond them: guaranteed return timing ('within seconds of a real change... or after maxWaitSeconds'), the split between in-progress results (only deltas) and terminal results (full picture), and what scores are absent (deductions, policy, coverage). The note that 'An area has no weight and feeds no other score' is a genuine semantic disclosure about the data model, not restated annotation text.

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 the change-event list are front-loaded, which is good, but the middle is one very long sentence stacking score taxonomy, exclusions and area weighting. Details such as 'each Report Area's Area Health score... for an area also the PageSpeed strategy' and 'An area has no weight and feeds no other score' sit awkwardly before the actually critical instruction to re-call with since, which is buried at the end.

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?

There is no output schema, so the description must carry the return contract, and it does: it enumerates the streaming message types (crawl milestones, category start/finish, heartbeat), the in-progress result fields (updates, changedCategories, run state), and the terminal result contents in both text and structured form. For a long-running, stream-oriented tool this is complete enough to call correctly.

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%, so the baseline is 3; the description earns above baseline by explaining the contract around since — that it must come from the previous watch_audit result and exists so 'categories that finished between calls are reported', which is causal meaning the schema only hints at. maxWaitSeconds and runId defaulting are largely restated from the schema, keeping it out of 5 territory.

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 — 'Follows an SEO Audit of a site' — and immediately adds the streaming-progress mechanism ('streams each change as a progress message') with enumerated event types. It distinguishes itself from siblings by describing a follow/watch semantic that differs from get_audit's snapshot, and it names start_audit as the predecessor.

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?

It gives explicit workflow guidance: 'Use it after start_audit' and 'call watch_audit again with the since value from this result until the status is completed, partial, failed or cancelled', which is exactly the polling protocol an agent needs. It falls short of a 5 because it never states when NOT to use this tool, e.g. that once the audit is terminal, get_audit should be used for the final structured read rather than another watch call.

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