Corollary — scientific AI models, tools and databases

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

Annotations already declare readOnlyHint=true and openWorldHint=false, covering safety and scope. The description adds what is returned (each model's function, licence, online availability, and selection guidance), but says nothing about permissions, rate limits, or data freshness, so it adds only modest behavioral context.

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?

A single, front-loaded sentence that moves from condition to return value to guidance. The list of examples is long but functional, though it could be trimmed without losing meaning.

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 tool with no output schema, the description explains what will be returned and that selection guidance is included, which is sufficient for correct invocation. It does not cover pagination or error cases, but those are minor for this simple comparison operation.

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 schema coverage at 100%, the schema already documents the single 'task' parameter. The description goes beyond the schema by listing concrete example tasks ('protein structure prediction', 'molecular docking', etc.), which helps the agent formulate valid values.

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: return every model or tool Corollary lists for a given task, including comparison guidance. It is clear what the tool produces, but it never names siblings like search_scientific_ai or get_scientific_ai_details, so differentiation is implicit rather than explicit.

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 opening 'For a task such as...' establishes the context: use when you have a task and need a comparison of all listed models/tools. However, it gives no when-not-to-use guidance and does not mention the sibling tools, so the agent must infer that search_scientific_ai or get_scientific_ai_details are alternatives.

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=true and openWorldHint=false, so safety is covered. The description adds genuine context by specifying the breadth of the returned reference (strengths, limitations, licence, and how to run it on Corollary), which goes 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?

Two sentences, front-loaded with the return scope and then the accepted input forms; little waste. Slightly dense enumeration of contents in one clause, but every item is meaningful to an agent.

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-required-parameter read tool with no output schema, the description must convey the return payload, and it does so by listing the reference sections. Nothing needed to call it correctly is missing, though the absence of output schema keeps it just short of exhaustive.

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 100% and the schema already says 'Name, id or Corollary page URL', so baseline is 3. The description earns an extra point by giving concrete accepted formats ('RFdiffusion', 'esmfold2', 'nf-core/rnaseq', 'facebook/esm2_t33_650M_UR50D'), disambiguating name vs id vs URL beyond the schema text.

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 resource ('Full reference for one entry') and enumerates what the entry covers (what it is, how it works, key facts, uses, strengths, limitations, citation, licence). The single-entry scope implicitly contrasts with the search/list/compare siblings, though 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?

Usage is only implied: telling the agent it must supply a name, id or URL signals this is a lookup used after obtaining an identifier, but the description never says when to choose it over search_scientific_ai or compare_scientific_ai_for_task, nor any prerequisite or exclusion.

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=true and openWorldHint=false, so safety is covered; the description adds real value by disclosing what comes back — ordered stages, programs/models/databases, common failure modes, and a starting page. It does not mention response size, ambiguity handling, or failures when a task is unmatched.

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 audience framing and the core promise before the example list, and the usage rule is a tight closing sentence. The eight-example list is long but each example broadens coverage of valid task domains, so it mostly earns its place.

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 one simple parameter and no output schema, the description carries the return-value burden well by specifying the shape of the answer. Minor gaps remain around what happens for unrecognized or overly broad tasks, but nothing essential is missing for correct invocation.

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 100% and there is only one parameter, so the baseline is 3. The description goes beyond the schema by illustrating the expected granularity and phrasing of 'task' with varied real examples ('fit an IC50', 'call variants'), which meaningfully guides input construction.

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 ('return how it is done on Corollary') and enumerates eight concrete task shapes, so an agent immediately knows the tool resolves a task description into a workflow. The closing clause distinguishes it from model-oriented siblings like get_scientific_ai_details.

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?

Explicitly scopes usage: 'Use this when the user wants a task done rather than a model named,' which cleanly separates it from the model-detail and comparison siblings. It stops short of naming a sibling tool by name or describing the inverse case, so it falls just below a 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?

Annotations already declare readOnlyHint=true and openWorldHint=false, so the safety profile is covered. The description adds that results include every supported analysis and its URL, but it does not describe pagination, ordering, or response shape.

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?

It is a single compact sentence with no filler. The wording is slightly awkward around 'method for, with its URL,' but it remains front-loaded enough for a simple listing tool.

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 zero-parameter, read-only list tool, the description covers the basic content. However, with no output schema, it does not fully explain the return shape or how this list differs from the sibling list_scientific_ai_tasks.

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 tool takes zero parameters, so there are no parameter semantics to explain. The baseline for a parameterless tool is 4, and the description appropriately does not invent parameter details.

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 identifies the resource as every scientific analysis Corollary supports and says each comes with a URL, so the agent can understand this is a catalog-listing tool. It does not distinguish itself from the closely named sibling list_scientific_ai_tasks, leaving ambiguity about which list to choose.

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 gives no explicit when-to-use guidance, prerequisites, or alternatives. With siblings such as list_scientific_ai_tasks and search_scientific_ai, the lack of routing guidance is a meaningful gap.

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=true and openWorldHint=false, so the safety profile is covered. The description adds one genuine structural fact beyond the annotations: each returned category includes a page URL. It says nothing about exhaustiveness, ordering, or whether the list is static, so it adds modest value only.

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?

A single compact sentence with no filler, and the return-content detail is front-loaded alongside the resource. It is a noun fragment rather than a complete sentence, which slightly weakens the structure.

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 no parameters and no output schema, the description carries the burden of describing the return shape, and it does state that each category comes with a page URL. It could note whether the list is exhaustive or stable, but for a zero-argument enumeration tool this is nearly complete.

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 tool takes zero parameters, so there is nothing for the description to disambiguate. Baseline 4 applies; no parameter-level gaps exist.

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 names the specific resource (the task categories Corollary compares models and tools in) and states what each entry carries (a page URL), so an agent can distinguish it from compare_scientific_ai_for_task and get_scientific_ai_details. It stops short of an explicit verb like 'lists', reading more as a definition of the data than of the operation, but the resource is unambiguous.

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?

There is no statement of when to call this versus the siblings — nothing like 'call this first to discover valid task identifiers before using compare_scientific_ai_for_task'. Usage must be inferred purely from the name and resource description.

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=true and openWorldHint=false, so safety is covered. The description goes further by disclosing what each result contains (purpose, licence, whether it runs online on Corollary with no GPU or installation, and the page URL), which is genuine behavioural context an agent cannot get from the annotations or schema. It does not mention ranking or result-count behaviour.

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 catalogue scope, then the result shape, then the usage trigger — a logical order with no filler sentences. It is longer than strictly necessary: the long parenthetical domain lists (protein structure prediction, protein design, docking, antibodies, genomics...) and tool/database name examples could be trimmed, though they do aid query formulation.

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?

There is no output schema, so the description commendably fills that gap by describing what each result reports (function, licence, online-runnability, page URL). Combined with the readOnly annotation, an agent has enough to call it correctly; only ranking/ordering and what happens on zero results remain unspecified.

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 100%, including an enum for 'kind' and an example-laden query description, so the schema carries the parameter semantics. The description's domain enumeration ('protein design, docking, genomics, single-cell, chemistry/ADMET' etc.) hints at useful query terms but adds no format or filtering guidance beyond the schema. Baseline 3 is correct here.

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 (search) and a precisely scoped resource (Corollary's catalogue of open scientific AI models, bioinformatics programs, databases, nf-core pipelines and datasets). The enumerated domains make it immediately distinguishable from siblings like compare_scientific_ai_for_task and get_scientific_ai_details without opening any schema.

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?

Explicitly states when to use it: 'when a user asks which model or tool to use for a biology or chemistry task, or where to run one.' It gives a clear situational trigger but never names the adjacent alternatives (compare_scientific_ai_for_task, get_scientific_ai_details) or says when to prefer those over a plain search, so routing between siblings still requires inference.

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