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

Annotations cover the safety profile (write, idempotent, destructive), so the bar is lower, yet the description still adds real behavior: the unknown-state fallback for non-proving evidence types and the return payload ('updated findings and report link'). It does not explain why destructiveHint=true for what reads as an append, nor that audit_key is the authorizing secret — those gaps keep it from a 5.

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 tight sentences, front-loaded with the core action and followed by the negative constraint, the edge-case rule, and the return. Efficient overall, though the 'sanitized observations... without payloads' clause partially restates the schema's evidence field description.

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, the description need not detail return values, and the brief return note is sufficient. It covers the append semantics, the source types, the fallback rule and the consent/time dimensions, leaving only the audit_key authorization prerequisite and the destructive-write consequence unexplained for an 8-param mutation.

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 50%: the write-control params (audit_id, revision, audit_key, request_id) are documented in-schema, while required params url, observed_at, consent_state and checks are bare. The description touches consent state, observation time and sanitization ('without payloads, identifiers or credentials'), but that sanitization note duplicates the schema's evidence description and adds no format or enum mapping.

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+resource ('Saves bounded runtime evidence to the same audit run') and enumerates the evidence sources (browser, backend inspection, destination UI). It differentiates functionally from browser-running siblings via 'Does not run a browser or send events', but never names the sibling tools it must be distinguished from (record_audit_context, submit_findings), leaving the routing to inference.

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 negative guidance ('Does not run a browser or send events') implies this is for supplying evidence already gathered rather than collecting it, which steers the agent away from audit_site/check_live_setup. However, there is no explicit when-to-use statement, no prerequisite for obtaining audit_key/revision first, and no named alternative for the collection case.

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