paper-preflight

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

Annotations declare readOnlyHint and idempotentHint, and the description reinforces this with the load-bearing detail that nothing is written and the agent must apply the diff itself. It further discloses what safe/unsafe actually mutate and the invariant that unfindable references are never fixed — context well beyond the 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?

Front-loads the core behavior and output format, then layers level semantics compactly. Sentences earn their place, though the level paragraph is dense enough to require careful reading.

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?

With an output schema present, return values need not be explained, and the description covers the write-safety semantics and level behavior well. The remaining gap is the un-annotated keys/path/offline parameters, which an agent must infer from names alone.

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 0%, so the description carries the burden. It fully explains the most consequential parameter (level, including the enum values' behavioral differences), but keys, path, and offline are left undocumented in both schema and description.

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 (propose edits) and resource (.bib files), plus the concrete output form (unified diff) and the source of the edits (verified records). This is clearly distinguishable from preflight_bib_lookup, preflight_check, and preflight_explain.

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?

Gives explicit guidance on level="safe" vs level="unsafe" and instructs showing unsafe diffs to the user before applying. It does not, however, route the agent against any named sibling (e.g. implying bib_lookup must run first), so alternative-selection guidance is strong but incomplete.

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

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

Annotations already declare readOnly/idempotent/openWorld, but the description goes well beyond them: it enumerates the four status values with their meanings, warns that a published preprint resolves to the published version with the eprint retained, and notes that status_flags can carry notices such as 'retracted'. These are exactly the edge behaviors an agent needs and cannot get from structured fields.

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?

Front-loads the core action and how to invoke it, then groups the outcome semantics into one tight paragraph. Dense but every clause carries information; no filler sentences.

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 an output schema exists, status/candidates/status_flags need not be re-explained, yet the description does so helpfully for the retraction and preprint-versioning cases. The only real omission is guidance on `prefer` and `offline`, which leaves an agent guessing about preprint preference and offline behavior.

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 full burden. It explains identifier, title, author, and year, but leaves `prefer` (published/preprint enum) and `offline` entirely unaddressed, so two of six parameters remain opaque. Partial compensation only.

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 concrete verb (get a BibTeX entry) and scope (for a DOI, arXiv ID, or title) and emphasizes it is built from the registry record rather than from memory. This is distinctive against siblings like preflight_bib_fix, which implies fixing an existing entry, so an agent can pick this one for retrieval.

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?

Gives clear invocation guidance ('Give `identifier`, or `title` (with `author` and `year` when you know them)') and prescribes the action for each outcome: use bibtex when found, disambiguate with the user or retry with author/year, and explicitly not to invent a source on not_found. It does not name sibling tools as alternatives, so it falls short of the full 5.

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

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

Beyond the annotations (readOnly/openWorld/idempotent), the description discloses pagination via offset/next_offset, the meaning of complete:false (a source was unavailable), offline cache-only behavior, and first-run latency of a minute or two with progress reporting. This is exactly the kind of behavioral context structured fields cannot carry.

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?

Front-loaded with the core purpose, then the return/pagination behavior, then latency caveats. Every sentence earns its place, though the paragraph is dense and could be broken up slightly for scanability.

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, yet the description helpfully still summarizes the return shape and the critical complete:false caveat. It is nearly complete for a read-only verification tool, with the only gap being the undocumented lang and include_info parameters.

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 compensate. It meaningfully explains max_findings, offset/next_offset, and offline, but says nothing about lang or include_info, leaving two of six parameters undocumented anywhere.

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 gives a specific verb and resource: verify every cited reference of a LaTeX project or .bib file against real scholarly records, plus check citation keys and the bibliography. This is clearly distinguishable from purely fixing or lookup tools, though it never names the sibling tools (preflight_bib_fix, preflight_bib_lookup) it should be contrasted with.

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 implies the usage context (verifying references before declaring a paper clean) and explains the pagination loop and offline mode, but never states when to choose this over preflight_bib_lookup or preflight_bib_fix, nor any explicit exclusions. Usage is inferable rather than spelled out.

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

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

Annotations cover the safety profile (readOnly, idempotent, openWorld), but the description adds genuinely new behavior: the first call downloads and caches the work's text from arXiv/OA/PDF or falls back to the abstract, and it enumerates evidence.level values ('full_text', 'abstract', 'none'). It stops short of describing cache staleness or rate/failure behavior on repeated calls.

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?

Front-loaded with what the tool returns, followed by concrete judging instructions that are actionable for an agent rather than filler. It is somewhat long, but nearly every sentence constrains agent behavior (quote verbatim, prefer 'could not confirm', never declare a citation wrong).

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 fields need not be re-explained, and the description adds the workflow and interpretive guidance needed for a non-trivial fetch-and-judge tool. The main completeness gap is the undocumented optional parameters, which the description does not address.

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 parameter burden, yet only `key` is meaningfully explained (the citation key being traced). `path`, `offline`, and `max_passages` are never mentioned, leaving their semantics to be guessed despite the low coverage.

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 precise verb+resource: for each citing sentence it returns the claim plus the best-ranked passages of the cited work. It is clearly distinguishable from siblings like preflight_bib_lookup or preflight_bib_fix, which operate on bibliography data rather than tracing a citation to source text.

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 implies the use case (deciding whether a cited work supports a claim) and gives a judgment protocol, but it never states when to choose this tool over preflight_check or preflight_explain, nor any exclusions. Usage is inferable rather than explicit.

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

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

Annotations already declare readOnlyHint, idempotentHint and openWorldHint=false, so the safety profile is covered. The description adds real context beyond that: the categories of information the caller will receive (what a rule detects, default severity, message template, fix applicability), which helps an agent decide whether this call answers its question.

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?

One tightly written sentence with the verb and resource front-loaded and the return payload enumerated after the colon. No filler or repetition.

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-only explainer with an output schema present, the description supplies everything needed to invoke it correctly and does not need to describe return values. Minor gap: no guidance on what happens for an unknown or malformed rule_id.

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 0% for the single rule_id parameter, so the description must compensate, and it does by giving concrete example identifiers (REF003, CIT001) that convey the expected format. It stops short of stating whether unknown IDs are rejected or aliased, hence not a 5.

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 ('Explain') and resource ('paper-preflight rule'), gives example IDs, and enumerates the content returned (detection, severity, message template, fix safety). This clearly separates it from sibling tools like preflight_check and preflight_bib_fix.

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?

Usage is implied by the read-only nature and the name, but the description never states when to call this instead of siblings (e.g., before running preflight_check, or to interpret a rule ID emitted by a check). No explicit when/when-not guidance or alternatives are named.

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