save_evals

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

The description reveals concrete behavioral properties beyond the annotations: unresolved kbids cause rejection, duplicates are dropped, and a single invented kbid fails the entire batch. These details are non-obvious and valuable for safe invocation, and they align with the idempotentHint and destructiveHint 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?

Three compact sentences communicate core purpose, validation behavior, and the critical invocation precondition with no redundant filler. The most important operational restriction is placed at the end for emphasis while the first sentence front-loads the primary purpose.

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 a single-parameter tool with a 100%-described schema and no output schema, the description covers the essential decision factors: persist semantics, validation rules, duplicate handling, and batch failure behavior. An agent has enough to call it correctly without hunting for hidden edge cases.

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?

The input schema fully describes the only parameter 'cases' as JSONL with query, expectedKbids, and grades. The description reinforces that kbids must come from actual search results, which adds some semantic color, but it does not add a substantive layer beyond the already-complete schema.

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 opens with a specific verb-resource pair ('Validate and persist retrieval eval cases') and identifies the source workflow ('drafted via the eval-generate prompt'). This clearly distinguishes the tool from the generic sibling tools like retrieve, learn, or search.

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 an explicit important constraint: call it only with kbids a search actually returned, and explains the consequence of violating that. However, it does not explicitly name an alternative tool or define a contrasting situation where a different tool should be used instead.

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