save_search

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

The annotations already declare readOnlyHint=false, openWorldHint=false and destructiveHint=false, so the safety profile is largely covered. The description adds real value on top: idempotency semantics for repeated request_id/data pairs and the explicit 'Local write only / never send an application' scope limit. It does not say whether an overwrite of an existing request_id changes stored data or how conflicts between different data under the same ID are handled.

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?

Four short sentences, front-loaded with the core action, followed by retry semantics, scope, and post-call instruction. Nothing is padded. The telegraphic phrasing ('Local write only.') is dense but each fragment carries distinct information.

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?

An output schema exists, so return values need no explanation, and the annotations cover the safety profile. For a single-parameter mutation tool the description covers idempotency, side-effect scope, and follow-up behavior adequately. Minor gaps remain around what happens on a request_id collision with differing payloads and where the saved record is subsequently visible.

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 description coverage is 0%, so the description must carry the load. It does explain request_id lifecycle (new per batch, safe retry) and characterizes report_markdown as the 'exact final' Markdown, but profile_version is only implied and the nested Recommendation fields (job_id, title, summary, draft, source_updated_label) get no coverage at all. Meaningful for the top level, incomplete for the nested structure.

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 states a specific verb and resource: saving selected recommendations plus the exact final Markdown along with the profile version. This clears up the ambiguity in the name 'save_search', which could suggest persisting a query rather than persisting recommendations. It does not explicitly contrast itself with siblings such as get_history, though it is the only write/persist tool among them.

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?

It gives concrete operational guidance: a new safe request_id per batch, and that an identical ID/data combination safely retries, which tells the agent how to behave on retry. It also imposes a hard boundary ('never send an application') and a post-success action ('confirm saving after success'). It stops short of stating the precondition for calling it, e.g. that recommendations must originate from a prior search.

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