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

Annotations are all false (not read-only, not idempotent, not destructive), so the description carries the disclosure burden. It adds meaningful behavioral context beyond the annotations: nothing is transmitted, the draft lands in the Drafts folder, and the owner acts on it externally. This clarifies the non-destructive save nature despite readOnlyHint=false, which is valuable context the annotations alone do not 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?

Three compact sentences with the core purpose and non-sending behavior front-loaded, followed by attachments and usage preference. No filler; each sentence earns its place. Slightly verbose in the attachments clause but well organized.

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 11-parameter tool with no output schema, the description covers purpose, safety, and usage routing well, but it never hints at the return value (e.g., whether a draft uid is returned for later send_draft calls), which an agent would need to chain operations. It is adequate but leaves the output contract to inference.

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 73% (moderate), so the schema handles most parameters. The description adds value by enumerating attachment sources ('existing mailbox attachments, files from upload_attachment/request_upload, or inline content') and implying reply context via 'preferred way to prepare replies.' Yet it adds little beyond the schema for cc/bcc/text/html/quote, and the reply preference is largely repeated by the in_reply_to_uid field description. It does not fully compensate for the 27% undocumented parameters.

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 ('Saves a draft into the Drafts folder'), and distinguishes itself from siblings by explicitly saying 'Nothing is sent' (vs send_draft) and 'This is the preferred way to prepare replies' (vs reply_draft/reply_send). The purpose is unambiguous and clearly separated from the 21 siblings.

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?

Provides clear context: it is the preferred way to prepare replies, and notes the owner will review/send it (optionally via send_draft). This routes the agent away from send_message/send_draft. However, it does not explicitly exclude reply_draft for direct replies — the in_reply_to_uid schema field carries that guidance ('Prefer reply_draft / reply_send') rather than the description, leaving a small ambiguity between create_draft and reply_draft.

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