get_final_metrics

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

With no annotations, the description carries the disclosure burden and does useful work: it names the three concrete metrics returned and warns that a run without an evaluation yields status 'no_eval'. This prevents an agent from misreading a sentinel as an error. It does not describe the response shape beyond the metric keys, keeping it short of a 5.

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?

Two tight sentences with the core action front-loaded and the edge-case behavior second. No redundant restatement of the tool name and nothing extraneous to trim.

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 one-parameter read tool with no output schema, the description covers the key return facts: which metrics come back and the 'no_eval' fallback. Slightly incomplete in that it doesn't indicate return structure (flat keys vs nested) or units, which matters when no output schema is published.

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?

There is a single required parameter at 100% schema coverage, and the schema already supplies the format and provenance ('exactly as returned by list_runs, e.g. lego_baseline'). The description's 'of one run' only reinforces what the schema documents, so the baseline 3 applies.

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 and resource ('Get the final test metrics') and enumerates exactly which metrics (psnr, ssim, lpips) for a single run. The word 'final' implicitly contrasts with get_metric_curve, but no sibling is named explicitly.

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 says what the tool returns but never says when to prefer it over get_metric_curve or compare_runs, nor any prerequisite (e.g. that an eval must have been run). The only condition mentioned, 'no_eval', is a return state rather than a usage rule.

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