nonlinear_icp

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

Annotations already declare `readOnlyHint: true`, so the safety profile is known. The description adds useful behavioral context: statistical assumptions (Causal Markov, faithfulness, acyclicity, sufficiency), preconditions for ICP, and failure symptoms like 'unstable skeleton / many undirected edges'. It does not describe the exact return payload, but the output schema covers that. This is a solid behavioral picture beyond the structured 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 dense but compact, covering alias identity, assumptions, preconditions, failure modes, alternatives, and minimum sample size in a few sentences. It front-loads the core identity and then structures supporting context logically. It could be slightly more readable with bullet-like separation, but every clause earns its place.

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 10-parameter causal discovery tool, the description covers the essential contextual ground: assumptions, preconditions, failure modes, alternatives, and sample-size guidance. The output schema and full parameter schema handle the mechanical details. Nothing critical is missing for an agent to decide whether to call this tool and how to interpret an unstable result.

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. The description adds meaning beyond the schema by explaining that ICP requires data labelled by environment/intervention, which directly clarifies the `environment` parameter, and by providing 'typical minimum N: 500' for the `X`/sample-size context. It does not individually explain `X` or `y`, but the schema already does.

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 identifies this as an alias for `icp(..., method='nonlinear')`, which clearly points to nonlinear invariant causal prediction, and it lists related sibling methods as alternatives. It does not state the core objective in a standalone verb+resource phrase, but it is unambiguous for an agent familiar with ICP. The distinction from `pc_algorithm`, `fci`, `ges`, and `lingam` is explicit.

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 names explicit alternatives (`sp.pc_algorithm`, `sp.fci`, `sp.ges`, `sp.lingam`) and gives decision-relevant context: ICP needs environment/intervention labels, while FCI is recommended if latent confounders are plausible. Failure-mode guidance ('Increase the sample, relax the CI-test threshold, or switch to FCI') tells the agent how to respond to bad results. This is strong when-to-use versus when-not-to-use guidance.

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