Detections

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

Annotations already cover the safety profile (readOnly, idempotent, non-destructive, closed world), so the bar is lower. The description adds real behavioral context beyond them: results are ordered newest first, 'compact drops the glossaries,' and the framing 'a detection is a shape with evidence, never a verdict' tells the agent how to interpret (and not over-trust) the output. It does not discuss volume, pagination beyond the limit cap, or latency.

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?

Two sentences, front-loaded with the resource and ordering, and every clause carries information (kind enumeration, evidence-not-verdict framing, filter surface). The dense comma list is slightly hard to parse but nothing is wasted or padded.

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 7-parameter, zero-required, no-output-schema read tool with solid annotations, the description supplies what is needed to call it correctly: what a detection is, what kinds exist, what ordering to expect, and what the filters mean. It leaves out one kind gloss and any notion of result shape fields, but nothing critical for correct invocation 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 only 57% schema description coverage, the description has to compensate, and it largely does: it glosses six of the seven kind enum values in plain language (drain_pattern, retry_storm, mint_burst, layering_fan, bind_closed_after_nft_transfer, god_key_unknown_7702) and explains that drain detections carry the share of wallet balance taken. It omits the uri_timeout_streak kind and says nothing about limit/severity semantics, so it is not exhaustive.

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 specifies the resource precisely — a log of anomalies emitted by detectors, ordered newest first — and enumerates the detection shapes (drain-shaped payments, retry storms, mint bursts, layering_fan, bind closures, unknown 7702 delegation), which lets an agent map prose to the kind enum without opening the schema. It never contrasts itself against siblings such as agent_pattern, blast_radius, or since_last, which could plausibly be confused with an anomaly feed, so it stops short of a 5.

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 states the filter surface ('Filter by agent, kind, severity, time or live incidents only') and the compact toggle, but those are restatements of schema parameters rather than guidance. There is no statement of when this tool is the right choice versus alternatives, nor any exclusion or prerequisite. Usage is only implied.

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