ml_bounds

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

The readOnlyHint=true annotation already covers safety, so the description's job is to add behavior beyond that — and it does: the output is an interval not a point, Lee/Oster variants impose different assumptions, and the failure mode (bounds too wide → add an auxiliary restriction) tells the agent how the tool's output can be uninformative and how to respond. Typical minimum N is also disclosed. It does not mention runtime or bootstrap behavior, but the output schema and schema-level parameter docs pick up some of that load.

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 compact (~75 words) with labeled sections (Assumptions, Pre-conditions, Failure modes, Alternatives) that make it scannable and front-loaded with the core purpose. Every chunk carries information. The telegraphic labels occasionally blur categories — 'For Lee bounds' and 'Oster's delta' read like assumptions before resolving into variant-specific notes — which is a minor structural cost.

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 17-parameter tool with a 100%-covered schema, an output schema, and read-only annotations, the description covers the essential non-schema ground: assumptions, data pre-conditions, failure modes, an alternative-routing list, and a sample-size heuristic. It is missing an explicit 'when not to use' statement and does not connect the tightening restrictions mentioned in failure modes to the named sibling tools, but nothing an agent needs to call it correctly is absent.

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 100%, so the baseline of 3 applies; every parameter is already documented in the schema, including the y_min/y_max tightness insight. The tool description adds context (Lee bounds require a binary selection indicator; typical minimum N=100) but never maps these to specific parameter names, so it supplements rather than replaces schema-level meaning. A 3 reflects that the schema does the heavy lifting and the description adds only marginal param-level value.

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 opening phrase 'ML-enhanced partial-identification bounds on the ATE' gives a specific verb+resource and sharply scopes the tool to set-identifying, ML-based interval estimates. The follow-up 'the result is an interval, not a point' distinguishes it from point-identifying ATE estimators. However, it stops short of explicitly contrasting itself with its closest siblings (e.g., 'for non-ML bounds use manski_bounds'), leaving that to the Alternatives line.

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 gives clear context: use when only weak set-identifying assumptions are credible, and it specifies pre-conditions (data for point-identifying analysis plus a weakest credible restriction, a binary selection indicator for Lee bounds). It lists alternatives (sp.oster_delta, sp.lee_bounds, sp.manski_bounds) and a failure-mode remedy. It does not state explicit when-not conditions or map each alternative to a concrete selection criterion, and the 'sp.' prefixes do not exactly match the sibling tool names.

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