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

Annotations already declare readOnlyHint=true, so the safe-read nature is covered. The description adds valuable behavioral context beyond that: the distinction between value changes and timestamp refreshes, the omission of withdrawn CVE IDs, and the cursor-based continuation mechanism. These are non-obvious and critical for correct invocation, and they align with the idempotentHint=false annotation (since cursor use makes the operation stateful). No contradiction with annotations.

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 a single dense paragraph that packs a lot of information, but it is not well structured—the list of event types is embedded in a long sentence, and the cursor usage instructions are tacked on. It is front-loaded with the main purpose but lacks clear paragraph breaks or bullet points. The content earns its place, but the presentation could be cleaner, so it does not merit a 4 or 5.

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?

Given the complexity of a change stream with pagination, event types, and CVE scoping, the description covers the essential usage flow well: initial call, continuation, and scoping. It also discloses edge behavior (withdrawn CVE IDs omitted). However, it omits details on the 'type' and 'limit' parameters and does not describe the output shape (since there's no output schema). For a read-only tool, this is reasonably complete, but the missing parameter explanations and output format leave some gaps, so a 4 is fair.

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 only 20% (only 'cve' has a description in the schema). The description compensates for three of five parameters: it explains 'since' (format and strictly-after semantics), 'cursor' (continuation via next_cursor), and 'cve' (scoping). However, it does not explain what 'type' or 'limit' do, nor does it explicitly state that 'type' filters by event type (though the event list is given). Given the low coverage, the description only partially compensates, so a 3 is appropriate.

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 is highly specific: it defines the tool as 'the publication change stream' and enumerates exactly which events it reports (first_published, kev_added, etc.), including the nuance that timestamp-only refreshes do not fire events. This clearly distinguishes it from sibling tools that fetch single records or current states, giving an agent a precise mental model of the tool's function.

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 provides clear operational guidance: pass 'since' on first call, then use the returned cursor to continue, and optionally scope by 'cve'. It implies the tool is for tracking changes over time, which differentiates it from sibling tools like get_cve or get_chains that likely return current snapshots. However, it does not explicitly state when NOT to use this tool or name alternatives, so it stops short of a 5.

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