Judgment Precision Report

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

Goes well beyond the provided annotations (readOnlyHint, idempotentHint, destructiveHint) by disclosing a genuinely subtle behavioral trait: decisions whose item.last_round equals the judge round AND no later round exists are excluded as 'no ground truth,' with an explanation of why ('we cannot tell if the ask would have come back'). This is exactly the kind of non-obvious scoping an agent needs to interpret results. Also adds version context and result shape. No contradiction with 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?

Substantial but every sentence earns its place: version marker, function, inputs, algorithm, output, exclusion rule, and consumer decision. The most important behavioral detail (the no-ground-truth exclusion) is front-staged where it can't be missed, and the dangling operational references ('v2.14.0', 'item 2 / v2.13') add provenance without bloat.

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 complex analytical tool doing multi-session aggregation with a non-trivial ground-truth determination rule, this description is remarkably self-sufficient: it covers inputs, mechanism, output contract, edge-case exclusion, and the downstream decision it serves. The only marginal gap is the lack of a defined return schema, but no output schema exists and the prose already enumerates the return shape (per-peer TP/FP/TN/FN plus precision/recall/F1).

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?

With 0% schema description coverage, the description carries full weight, and it largely delivers: it adds meaning to three of the four parameters by embedding them in the algorithm ('via session_id, or filtered by judge peer / since timestamp'), clarifying that 'peer' means the judging peer and 'since' is a timestamp filter. response_format is left undiscussed, though its name and default make it self-explanatory. Strong compensation for the schema's silence, though not exhaustive.

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?

Opens with a specific verb+resource: 'compute precision/recall/F1 of the shadow judge against the empirical ground truth,' then details the exact event stream, correlation mechanism, and output shape. The extreme specificity of the resource ('session.evidence_judge_pass.shadow_decision' events vs. 'evidence_checklist resurfacing behavior') makes confusion with siblings like session_metrics or session_peer_reliability_report implausible, even without naming them 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?

Provides a concrete 'when to use' statement: 'Operator uses this to decide whether to flip a peer from shadow to active mode (item 2 / v2.13),' which ties the tool to a real decision gate. However, it never names the overlapping alternatives (session_peer_reliability_report, session_metrics, session_doctor) nor states when NOT to use it, leaving the exclusion side implicit.

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