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

Annotations declare readOnlyHint=false, destructiveHint=false, idempotentHint=false, openWorldHint=false, so the safety profile is already structured. The description adds genuinely non-structured behavior: a two-phase deferred write that returns a pending_write_id, a mandatory human confirmation gate in the same conversation, and output-contract rules (attribution footer must be reproduced verbatim; only returned fields count as verified). These are real obligations an agent could not derive from the schema or annotations.

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 critical never-saves-immediately constraint are front-loaded correctly, which is the right structure for a write gate. However, a large fraction of the text is generic data-accuracy and attribution boilerplate (no invented budgets, ROAS, CPA, 'waste' labels) that reads like a global policy rather than tool-specific guidance, diluting the definition.

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 does the necessary work of stating what the call returns (a pending_write_id) and the exact downstream step needed to persist the fact, plus the confirmation precondition. The remaining gap is the absence of any field-level semantics for the three required inputs.

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?

All three parameters (key, value, category) are required and schema description coverage is 0%, so the description carries the full burden — and it never explains what key/value/category should contain, what formats are valid, or how category is constrained (it has a default of 'general' but no guidance). It refers only to 'the exact fact', which does not map to the three fields.

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 (Prepare a write) plus resource (a declared canonical fact) and immediately distinguishes itself from siblings by declaring it does not save immediately and by naming canonical_pending_commit as the follow-up. An agent can tell it apart from canonical_facts_get/list, canonical_pending_commit, and canonical_pending_cancel without opening any schema.

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 lifecycle instruction: call only after proposing the exact fact to the user, then call canonical_pending_commit with the pending_write_id once the user explicitly answers yes in the same conversation. It also states an exclusion ('Never silently save inferred or guessed facts') and names the alternative tool, leaving nothing to inference.

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