Merron

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

Annotations already declare readOnlyHint=true, idempotentHint=true, and destructiveHint=false, so the safety profile is covered. The description adds genuine behavioral context beyond the annotations: the tool does not classify, does not treat omitted answers as 'No', and requires no sign-in/private access. These are meaningful disclosures that help the agent set expectations.

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?

Four sentences, each earning its place: purpose, parameter handling, behavioral limits, and access constraints. The purpose is front-loaded in the first clause, and there is zero filler or repetition of annotation information.

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-optional-parameter read-only tool with safety annotations already provided, the description covers purpose, returns ('questions and explanations'), parameter semantics, and access limits. Minor gaps remain: the description never explains how the 'selected AI system' is established (e.g., prior session state or sibling tool output), and the fact keys are 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 0%, so the description must compensate, and it does: 'Pass only known user answers; omit unknown values or use null' directly explains what the 'facts' object represents and how to populate it, aligning with the schema's allowance for null. It does not enumerate the fact keys, but it gives the parameter real semantic meaning beyond the raw schema.

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 states a specific verb and resource: 'Get relevant business questions and explanations for a selected AI system.' It also delimits scope with 'Does not classify,' which implicitly separates it from the screening sibling. However, it never names the sibling tools (merron_scan_project, merron_screen_ai_act) explicitly, so the differentiation is inferential 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 description gives clear invocation guidance ('Pass only known user answers; omit unknown values or use null') and a constraint ('No sign-in or private workspace access'), but it never states when to choose this tool over its siblings or when not to use it. The when-to-use context is implied by the tool name and the 'does not classify' remark rather than stated.

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 (which only declare non-readonly, non-destructive, non-idempotent), the description discloses the return artifact (private single-use link, valid 7 days), the downstream sign-in and prefilled-workspace flow, and hard data-handling constraints (never include code excerpts, secrets or personal data; findings carry only paths, line numbers and codes). That is exactly the extra behavioral context the annotations cannot supply.

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 purpose, then the consent requirement, then the data-safety constraint and link-handling instruction, in a sensible priority order. It is a dense paragraph with several clauses, but each sentence carries operational value; no obvious filler.

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 write tool with no output schema, nested object parameters and only half-covered schema descriptions, the description still covers purpose, consent, return value (link + validity), downstream flow and privacy boundaries. Nothing an agent needs in order to call it correctly is missing.

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 only 50% (only userAgreed is documented), so the description must compensate, and it does: it spells out the payload contents — name, description, confirmed answers and file/line findings from merron_scan_project — which maps to the nested system fields. It stops short of explaining the facts/company/productName fields in detail, so 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?

The description opens with a specific verb+resource ('Save this AI system to Merron so the user can build an evidence pack from it') and enumerates exactly what is persisted. It is clearly distinguishable from siblings: merron_scan_project produces the findings this tool consumes, and merron_get_questions supplies the confirmed answers.

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 tells the agent when to invoke it (after the user confirms answers and scan findings exist) and imposes an explicit precondition: 'Ask the user first and tell them what will be saved.' It does not name a when-not case or a competing alternative, so it falls short of 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 readOnly, idempotent, closed-world and non-destructive, but the description adds substantial behavior beyond them: a mandatory consent gate before data leaves the user's environment, and explicit non-capabilities (no URL fetching, no code execution, no input persistence, no access to private cases). This is exactly the extra context annotations cannot convey.

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?

Three sentences, all front-loaded: purpose and return shape first, then the consent prerequisite, then the prohibitions. Dense and largely waste-free, though the guardrail sentence packs four distinct exclusions into a single long clause, slightly reducing scannability.

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 output schema, the description carries the return contract and does so ('file/line references and follow-up questions'). Safety semantics, consent flow and scope limits are all covered; the only omission is any hint about the 12-file / 8,000-character input caps, which the schema does enforce on its own.

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 50%: sharingConfirmed is documented in the schema, while the files array (path/text/startLine, max 12 items, 8000-char text) has no per-property description. The description reinforces the consent semantics of sharingConfirmed ('obtain the user's agreement', 'user-selected excerpts') but adds nothing about the files payload shape or limits, so baseline 3 is appropriate.

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 ('Find') and precise target artifacts (AI integrations, model calls, disclosure wording, sensitive-use wording) scoped to 'user-selected project excerpts.' The clause 'Returns file/line references and follow-up questions, not a legal classification' explicitly distinguishes it from the sibling merron_screen_ai_act, which is the classification tool.

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 an explicit precondition ('First explain that the excerpts will be sent to Merron and obtain the user's agreement') plus hard boundaries on what not to send (secrets, personal customer records, whole repositories, unauthorised files). It does not explicitly state when to prefer this over merron_screen_ai_act, so it falls short of the full when/when-not/alternatives bar.

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 destructiveHint. The description adds substantial behavioral context beyond those: it states the computation does not check documents, use an AI model, save a report, access accounts, or certify compliance. It also imposes a strict rule to never infer confirmation or legal compliance from source code or absence of signals, and clarifies that unknown answers remain unknown. This is rich, non-redundant transparency.

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 three sentences, each carrying distinct weight: the first states purpose and outputs, the second sets constraints on inference and unknown handling, the third lists exclusions. It is front-loaded with the core purpose and does not waste words. Every sentence 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?

For a read-only, idempotent screening tool with two parameters (one nested object) and no output schema, the description covers the operation, return categories, constraints, and exclusions. It does not explain the exact output format, but the listed components give sufficient guidance, and the schema documents parameter types. Nothing critical is missing for an agent to call it correctly.

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 50%: factsConfirmedByUser has a description in the schema, while facts does not. The tool description compensates by explaining that facts must be explicitly confirmed by the user and that unknown answers remain unknown, adding semantic meaning beyond the schema. However, it does not detail the structure of the facts object (e.g., allowed values), which the schema partially provides. Overall, the description adds meaningful value over the schema.

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 clearly states the action: run Merron's versioned, bounded AI Act rules on explicitly confirmed facts, and lists the specific outputs (conditional screening result, unresolved routes, missing questions, official references). It also explicitly enumerates what it does not do, which distinguishes it from siblings like scan_project. This is a specific verb+resource statement.

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 strongly implies the tool is only for facts the user has explicitly confirmed, and warns against inferring confirmation from code or absence of signals. It also clarifies that it does not check documents, use an AI model, or certify compliance, which helps an agent choose this over alternatives. However, it does not explicitly name sibling tools or provide a direct 'use this when...' vs 'use that when...' comparison, so it is clear but not fully explicit.

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