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

Annotations only declare the generic write profile (readOnlyHint=false, non-idempotent, non-destructive). The description adds substantial context the annotations cannot: new rows land status='unapproved' pending human review, approval is recorded under the user's name as "settled without review", is reversible in one click, and is refused with 403 unless the caller is an owner or owner-delegated approver. That is exactly the operational detail an agent needs before mutating graph state.

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?

Front-loaded with purpose and workflow position before the mechanics, and each paragraph covers a distinct concern (params, approve semantics, response handling). The approve paragraph is somewhat verbose and repeats the consequence framing, but the risk it mitigates justifies most of the length.

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?

Despite a nested-object `results` parameter and no schema descriptions, the description covers inputs, workflow placement, approval authorization, and the two meaningful response fields (validation_errors, near_duplicates) even though an output schema exists. An agent has everything needed to call this correctly and handle its result.

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 0%, so the description carries the full burden and does: `memory` is the graph's slug, `results` maps each pass `key` to that pass's JSON with a concrete example, `source_doc` is an optional provenance label, and `approve` has both a default (false) and a strict conditional trigger. Format and intent are supplied for every parameter.

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+resource ("Save the extraction you produced") and immediately positions it in the workflow as the LAST step after get_extraction_prompt and resolve_names, explicitly naming two siblings. An agent can place this tool in the pipeline 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?

Gives explicit sequencing (after get_extraction_prompt and your single resolve_names call), an exclusion rule for approve ("ONLY when the user explicitly asked... Never set it on your own judgement"), and routing for response anomalies (report validation_errors, fold near_duplicates with merge_objects). Nothing is left to inference.

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