eval_toxicity

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

No annotations are provided, so the description carries full burden. It describes the scoring logic (four yes/no questions, fraction), the returned output schema, and that it evaluates LLM output. It does not mention any destructive side effects, which is appropriate for a read-only evaluation tool.

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 concisely structured with a clear one-line purpose, followed by grading details, usage guidance, and parameter explanations. Every sentence adds necessary information without redundancy.

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 the tool's simplicity (2 parameters, output schema provided), the description covers all necessary aspects: purpose, scoring method, usage context, parameter details, and return value structure. No gaps remain.

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%, but the description adds full meaning: 'output' is defined as 'The LLM-generated text to evaluate' and 'judge_model' is explained with examples and default. This compensates entirely for the schema's lack of descriptions.

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 detects harmful content in LLM output, with a specific verb 'detect' and resource. It explains the QAG-grading mechanism, distinguishing it from sibling eval tools like eval_bias or eval_hallucination.

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 explicitly says to use for guardrails, chatbot turns, or user-facing LLM output where harmful content is a risk. It provides clear context but does not mention when not to use or compare directly to alternatives.

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