Cross-check structured findings for contradictions

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

With no annotations provided, the description carries the full burden — and discharges it completely. It discloses the input boundary (server never reads the material), the output framing (every finding phrased as a possibility and paired with item numbers), the heuristics (rounding-explained differences downgraded, same-source conflicts specially noted, start-after-end date ordering weighted high), unreadable-value handling with reasons, and the zero-result contract that explicitly disclaims correctness and lists human-only lenses. Nothing about the tool's behavior is left to guesswork.

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?

The description is long — roughly 400 words across the bilingual text — but the tool's complexity justifies most of it. The core purpose and the critical constraint (server never reads the document) are front-loaded, and the seven checks follow a consistent, scannable structure; the bilingual duplication and granular parentheticals cost some conciseness but serve the ja/en audience and encode genuinely useful heuristics.

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?

Given no output schema, the description explains the return contract thoroughly: numbered findings referencing both item numbers, weak vs. high-weighted date-order findings, unreadable-value reports with reasons, and an explicit zero-result message with human-review lenses. The seven-check enumeration plus edge-case handling (rounding, same-source, unit ambiguity) covers the full behavioral surface of a high-complexity tool; the only unspecified detail is the exact JSON shape of the output, which the semantic detail renders non-critical.

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 high, so the baseline is 3, but the description adds meaning the schema lacks — most notably for `value`, which has no schema description at all: it specifies how values are interpreted (numeric with Japanese magnitude/qualifier/rounding parsing, date reading for date-like strings) and what happens when parsing fails. It also ties `subject` and `confidence=inferred` to specific checks, connecting each parameter to the behavior that consumes it.

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 title and opening sentence name a specific verb and resource — mechanically cross-checking the list of items you structured after reading material — and enumerate precisely the seven checks performed. It is self-evidently distinct from all 70+ siblings: no other tool concerns contradiction checking of structured findings, and the 'server never reads the document' clause separates it from the document-reading siblings such as read_document and extract_from_document.

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 clearly states the trigger context: use it on the items you structured after reading material, and the opening framing implies the tool is for post-reading verification rather than document access. The 'server never reads the document' clause is an implicit when-not, but no alternative tools are named, and the boundary against validation siblings like check_capability_map or analyze_text_with_claude is left for the agent to infer.

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