email_draft_create

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

No annotations are provided, so the description carries the full behavioral burden. It does disclose two important behaviors: creation is idempotent when the key is reused, and retries return the 'durable prior result.' However, it does not describe the return value shape, whether this creates a provider-side draft vs. local, rate limits, or what happens on failure. Given zero annotation coverage, the description gives modest context but misses the opportunity to explain what the tool returns or persists.

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?

The description is a two-sentence, tightly written block with no wasted words. It front-loads the primary action ('Create exactly one provider draft') and packs essential constraints (idempotency, retry behavior) into the second sentence. It could arguably be slightly more comprehensive given the tool's complexity, but every existing sentence earns its place.

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?

Given the tool's moderate complexity (6 params, 3 attachment shape variants, no output schema, no annotations), the description covers idempotency behavior and attachment modes but leaves gaps: it doesn't explain the return value, whether email addresses require validation, or how drafts interact with email_draft_list for retrieval. The complexity warrants more documentation than is provided, though the description covers the most critical idempotency constraint.

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 only 17%, so the description must compensate, and it partially does by explaining the three attachment shape options (retained, staged, confirmed local) and the critical idempotencyKey semantics ('reuse it only for the identical draft'). The description adds meaning to idempotencyKey beyond the schema's 'Required nonblank retry key.' However, it doesn't add semantic depth to parameters like accountId, to, subject, or body, which remain schema-only. The description partially compensates for the low schema coverage where it matters most.

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?

Description clearly states the tool's function: 'Create exactly one provider draft' with the specific scoping 'exactly one' and optional attachment handling. It identifies the resource (draft) and the verb (create), and the idempotencyKey requirement distinguishes this tool's behavior from siblings like email_send or email_draft_list. The precision of 'exactly one provider draft' conveys that this creates a draft at the provider level, which distinguishes it from local staging operations.

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?

The description provides explicit usage guidance: 'A nonblank idempotencyKey is required; reuse it only for the identical draft so retries return the durable prior result without creating a duplicate.' This clearly tells the agent when to use the tool (to create a draft) and the critical constraint (reuse key only for identical drafts to leverage retry semantics safely). It also defines the attachment modes (retained, staged, or confirmed local) which sets up the when/which selection criteria against sibling attachment-stage tools.

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