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

Annotations already cover the safety profile (readOnly, non-destructive, closed-world), so the description's burden is lower, yet it adds a great deal: the statistical method (k-let-preserving shuffles, FDR controlled within each statistic), the explicit absence of thresholds, and a concrete cost model ('roughly a third of a second per map per shuffle', 'minutes of work' at 99 shuffles). It also discloses the result shape (per-statistic rate with exact binomial interval plus per-map tests), which is exactly the kind of behavioral context annotations cannot carry.

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?

Purpose is front-loaded in the first sentence and the remaining sentences each carry load (justification for the statistical approach, sibling routing, cost model, cost-reduction lever). It is dense and the opening sentence is a run-on with an embedded contrast, so the structure costs some readability, but for a 10-parameter statistical tool the length is defensible rather than padded.

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?

An output schema exists, so return values need not be explained, and the description still sketches the result anyway. For a read-only, closed-world tool with nested `_subject`/`_files` inputs, it supplies the method, the no-threshold rationale, the sibling that handles thresholds, and the compute budget — everything an agent needs to decide and call correctly.

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 the baseline is 3 and the schema already documents every field including defaults, ranges and the 1/(shuffles+1) p-value bound. The description goes beyond it by tying parameters to operational consequences: `flags` controls cost, `shuffles` bounds the achievable corrected q, and the three statistics models map onto the flag set. That cross-parameter interaction is genuine added meaning, though no syntax or format detail is added.

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 (measure/discover the rate of) on a specific resource (cryptic σ70 promoters, splice sites, synthesis-breaking repeats across a corpus), and immediately distinguishes itself from raw hit counting. It also names the sibling it is not: threshold/repeat-limit questions are routed to `check_manufacturability`. An agent can separate this from `scan_sigma70_promoters` or `scan_splice_sites` without opening either schema.

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 a clear when-not (anything asking for a threshold is out of scope, and `check_manufacturability` answers the vendor-limit question) and a cost-control hint ('drop `flags` to the statistics you need to make it cheaper'). The core when-to-use — you want significance against a null rather than a count — is stated as a contrast rather than a directive, and the raw-count siblings are not named explicitly.

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