dataset_statistic

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

Annotations only declare readOnlyHint=true; the description adds the important trait beyond that: per-feature full results are not returned but retained in the session. It also discloses numeric-contract behavior (effect size = log2 of pre-transform arithmetic mean ratio, no pseudocount, BH population limited to testable features). It does not say how the session-held results are later retrieved, nor how long they persist.

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 the core action, then the two parameters, then the differentiators. Dense but every block earns its place; the v1 comparison paragraph is long but is the main disambiguation content. Minor redundancy with the indentation/line-break formatting that adds lines without adding meaning.

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 read-only, no-output-schema, two-parameter tool with an untyped nested object, this covers what an agent needs: what runs, which matrix it runs on, why the matrix must be explicit, how the numbers differ from v1, and where the full results go. The only residual gap is retrieval of the session-stored results, which is arguably another tool's concern.

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% and the nested `specification` object is an open additionalProperties object, so the description carries the full burden — and it does: it names the shared keys (statistic_id / kind / transform / feature_scope) and the kind-specific keys (reference_group+test_group, groups+alpha, scaling+n_components), and ties `matrix_result_id` to `analysis-matrix.v1` artifacts produced by pipeline preprocess / qc_processed.

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 with the exact statistic families enumerated (welch / anova_tukey / pca) and the scope constraint that it runs against an explicitly named analysis matrix. It also explicitly contrasts itself with the sibling dataset_differential, so an agent can pick between them without opening a 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 concrete usage rule — never implicitly reuse the most recent preprocessing, always name the matrix_result_id, because two matrices from different recipes may coexist. It explains how this differs numerically from v1's dataset_differential, but stops short of an explicit 'use this when X, use dataset_differential when Y' routing statement.

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