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

Annotations already establish read-only/idempotent/non-destructive, and the description adds material context beyond them: cursor-pagination protocol (feed next_cursor back as since), a polling cadence, an explicit cost statement ('unmetered in v1'), and the negative guarantee that no-op transitions are suppressed. That is exactly the delta the annotations cannot convey.

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?

One tight paragraph, front-loaded with the identity and event vocabulary, then pagination, then filtering, then cost/cadence, closing with a disambiguating negative. The telegraphed style is dense but every clause carries information an agent needs.

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?

With no output schema, the description does the work of describing what a record represents (observed transitions) and the vocabulary of event values. It leaves the per-event payload shape and the `limit` cap implications unstated, a modest gap for a read-only paginated endpoint.

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 80%, so the schema carries most parameter documentation. The description still adds real value by enumerating the accepted event names — the `events` array has no enum in the schema — and by restating the cursor-round-trip loop, which is the easiest thing for an agent to get wrong.

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?

Names the specific resource (a server-transition change feed) and enumerates the exact transition types it emits, so an agent knows precisely what it will get back. It also draws a boundary against a neighboring concept ('Not a malware scan'), making it distinguishable from siblings like census_lookalikes or census_search.

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?

Gives clear operating context: poll hourly, unmetered in v1, only real transitions surface (never 'still fine'), and filter along events/server_names/namespace_domain. It stops short of naming which sibling to use instead (e.g. census_recent vs census_watch_subscribe), so it is a strong 4 rather than a full 5.

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