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

Annotations only flag that this is a non-read-only, non-idempotent write, so the description carries real weight and delivers: it spends workspace credits, costs a full model charge or one credit on cache hit, returns a request rather than a result, and shows nothing to the user. The async/polling contract is disclosed well. Minor gap: it does not say what happens on a second identical call beyond the credit note, leaving the non-idempotent hint's practical impact slightly ambiguous.

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 exactly-one-of constraint are front-loaded, followed by cost, async lifecycle, and display handoff in that order. Every sentence carries operational information an agent would otherwise have to guess; nothing is redundant with the title or schema.

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 an async, credit-spending job with no output schema, the description covers the full lifecycle: what is charged, that the return is a request id, how to poll, how to read the grade, and how to surface it via sightglass_show. An agent can execute the whole flow from this text alone.

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 baseline is 3, and the description earns an increment by adding a mutual-exclusivity rule ('Pass exactly one of adId or sitePageId') that the schema itself does not encode via oneOf. It does not clarify projectId vs workspaceId precedence, but the schema already describes both.

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?

States a specific verb and resource ('graded teardown of one ad... or one landing page, scored against what works in its category') and precisely bounds the input domain. It is clearly distinguishable from siblings like sightglass_get_ad or sightglass_find_ads, which only retrieve rather than grade.

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?

Explicitly gates invocation: 'only call it when the user asked for a grade', plus the credit-cost rationale. It also names the follow-on tools and their conditions (poll sightglass_get_xray, then call sightglass_show once with resultId), so the agent knows both when to start and what to do next.

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