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

Annotations already declare readOnlyHint=false, destructiveHint=false, and openWorldHint=false, so the description's job is narrower, and it adds genuinely useful behavior: 'the proposal stays inactive until approval' and the required human-in-the-loop plus workflow_dry_run sequence. It does not say whether a second call replaces a pending proposal or how to reference the proposal afterward, which are the main remaining behavioral unknowns.

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 purpose, then usage, then sequencing, then a limit, then a compact JSON example — a well-ordered and dense definition with no filler sentences. The example is the largest block but earns its place by showing the nested payload an agent must construct.

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 non-read-only mutation with a nested, undocumented schema and no output schema, the description covers the workflow position, approval gate, and next step well. It still leaves an agent guessing about the metric-type semantics and about whether the response returns an identifier needed to trigger approval, which matters given 0% schema coverage.

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% across a nested object, so the description carries the burden, and the inline example does anchor the shape (workflowId, mainMetric.type, mainMetric.scope, guardrails array). However, it never explains what the enum values (judge_score, field_accuracy, human_edit_rate, cost, latency) measure, nor what nonInferiorityMargin, minimum/maximum, or nodeId mean for a node-scoped metric, so meaning is added only structurally.

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: 'Propose one main metric and guardrails for a workflow,' which tells an agent exactly what is created. It also draws a boundary against approval workflows ('Do not use when you need to approve or change approved checks'), though it never names the sibling tool that performs approval, so differentiation is by scope rather than by tool name.

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 trigger ('Use when an update needs quality checks') and an explicit exclusion ('Do not use when you need to approve or change approved checks'), plus a sequencing instruction (get human approval, then call workflow_dry_run). The only gap is that the excluded case does not point to a concrete alternative tool such as workflow_guard or judge_propose.

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