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

The description adds substantial behavioral context beyond the annotations: it specifies 'Pure pattern matching, no AI call,' meaning deterministic and cheap; it discloses 'Best-effort, not a guarantee,' warning about false negatives; it mentions the cost '(1 MESH/call)'; and it describes the return shape ({scrubbed, redacted, kinds}) despite no output schema. This goes well beyond the sparse annotations (readOnlyHint:false, openWorldHint:true, destructiveHint:false) and provides key transparency for an agent.

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 efficiently structured: it opens with a purpose clause, lists covered data types, gives a usage hint, describes input/output, warns about limitations, and notes cost. It is not overly long, but it begins with 'Safety Scrub —' which repeats the title, and the parenthetical '(1 MESH/call, a tool · safety)' is metadata that could be omitted without loss. Still, every substantive 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?

The tool has a nested input object and no output schema, so the description carries significant responsibility. It fully covers the return format ({scrubbed, redacted, kinds}) and even gives an example ('card','ssn'). It also covers behavioral nuances (best-effort, no AI, cost). Minor gaps exist: it doesn't explain aliases (though schema covers them) or behavior when no sensitive data is found (though 'redacted: boolean' implies it). Overall, it is sufficiently complete for the agent to invoke and interpret results.

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 input schema has 100% description coverage: the text parameter is described as 'REQUIRED — the text to scan; card numbers, SSNs, API keys/tokens, JWTs and PEM private keys are replaced with [REDACTED-*] markers.' The tool description only says 'Input: {text: string},' adding no new information beyond the schema. It does not mention the alias parameters (input, content) or the fallback chain, but since the schema documents those, the description doesn't need to. 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?

The description clearly states the tool's function: 'Redact sensitive data from text before logging it or sending it to a model.' It specifies the exact data types handled (cards, SSNs, API keys, JWTs, PEM keys) and the output markers ([REDACTED-*]). This distinguishes it from sibling tools, none of which mention redaction or safety scrubbing, and explicitly notes 'Pure pattern matching, no AI call' to differentiate from AI-based tools.

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 a clear usage directive: 'Use whenever user-supplied or scraped text may carry credentials.' It also implies when to use by mentioning 'before logging it or sending it to a model.' However, it does not explicitly name alternative tools or state when not to use it, so it falls short of the highest bar for usage guidance.

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