Do two metrics move together?

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

Annotations already mark the tool as read-only, idempotent, and non-destructive. The description adds substantial behavioral detail beyond those hints: it explains the detrending method (removing each weekday's usual level), the autocorrelation correction to the effective pair count, and the practical note that under ~3 months the answer is typically 'cannot_tell'. It also reveals that `first_pair` exposes lag misalignment, giving the agent a way to sanity-check its lag choice. No contradiction with annotations.

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 fairly long (several sentences) but every sentence earns its place: the core question, the methodological caveat, the output outcomes, and the lag explanation. The front-loaded question gives immediate purpose. It is not rambling or redundant, though a slightly tighter phrasing could improve it.

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 explains the possible return values and the data sufficiency threshold. It covers the lag alignment, the interpretation caveat, and the autocorrelation correction. For a complex statistical tool, this is reasonably complete; the only omission is a precise description of the response structure (e.g., fields like `first_pair` and the verdict), but that is not strictly required without an output schema.

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 100%, so every parameter is described. The description adds value beyond the schema by explaining the domain-specific alignment of sleep and readiness under the waking day (lag=1 example) and warning against testing multiple lags. It also reinforces the start/end length requirement. This goes beyond dry schema definitions and genuinely helps an agent choose and interpret parameters correctly.

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 precise question, 'Do the day-to-day changes of two metrics move together, beyond chance?', identifies the resource (two time series) and explicitly names the sibling tool `oura_compare` while distinguishing its method. It also lists the three possible outcomes (outside_noise, within_noise, cannot_tell), making the tool's role unambiguous and immediately differentiable from its siblings.

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 mentions `oura_compare` and states they share a rule, and notes that it measures co-movement not causation. However, it never explicitly states when to prefer `oura_relate` over `oura_compare` or other siblings, nor does it give exclusions like 'use compare for cause, relate for correlation'. The context is present but the guidance lacks an explicit decision rule, so it is only partially helpful.

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