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

Annotations (readOnlyHint, openWorldHint, idempotentHint) already establish safety, so the description's job is to add behavior, and it does so extensively: substring matching semantics with a concrete false-positive example, delivery-date (not Date: header) matching with whole-day precision, explicit cap disclosure with true match totals, and the warning that UIDs are only meaningful within their folder. These are exactly the failure modes an agent would otherwise discover by making mistakes.

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?

Longer than average, but every sentence carries operational weight — there is no filler. The structure is logical: purpose and output contract first, then the cross-folder guidance, then matching gotchas, then cap behavior. The caps-lock sentence is an intentional emphasis device for the single most cost-relevant behavior. It earns its length given the density of non-obvious semantics.

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 correctly carries the return-shape burden via a reference to sibling list_emails, and covers ordering, cap disclosure, and the empty-search refusal. The boolean filters (seen/flagged/answered) are left to the schema, which describes them adequately. A minor gap is that criteria combination semantics (AND vs OR) is never stated, which an agent might guess wrong with multiple criteria.

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, and the description adds genuine collective meaning: it unifies the substring-matching rule across from/to/cc/subject/body/text and the delivery-date rule across since/before/on, which the individual schema entries don't convey as an integrated semantic. The mailboxes one-call-cost point is also reinforced in prose. The only thing not added is how multiple criteria combine (AND vs OR), but the description clearly lifts above the schema baseline.

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 — 'Search one or more IMAP folders and return summaries' — and precisely defines the output contract: the shape list_emails returns, plus the folder each result was found in, newest first, not message bodies. It explicitly names read_email as the alternative for full message bodies, making sibling differentiation immediate.

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?

Explicit routing guidance is present: read_email is named as the tool for reading one message in full, and list_mailboxes (in the schema) is named for unknown folder names. The SHOUTED instruction to search every candidate folder in one call, with the concrete 'where is that email from Bob' example covering INBOX/Archive/Sent, gives the agent a direct decision rule tied to user intent. It also states the empty-search refusal behavior.

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