fepois

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

Beyond the readOnlyHint/openWorldHint annotations, the description discloses the statistical assumptions (exponential conditional mean, strict exogeneity/PPML consistency), two specific failure modes (convergence failure and separation) with step-by-step remedies, a validation parity claim, and a minimum-N heuristic. This is exactly the behavioral context an agent needs to anticipate and handle estimation failures.

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 purpose is front-loaded and sections are labeled (Validation, Assumptions, Pre-conditions, Failure modes, Alternatives), which aids parsing. However, the description runs roughly 280 words and includes vague entries like 'certified parity evidence' (parity with what?) and 'with caveats' (unspecified), and the alternative names use an 'sp.' prefix inconsistent with the tool registry — these ambiguities cost it a point.

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 21-parameter tool with a rich output schema and fully documented parameters, the description covers the important meta-context: assumptions, pre-conditions, failure modes and remedies, alternatives, and sample-size guidance. The remaining gaps — e.g., pairing vcov choices with failure scenarios or interpreting the detail levels — are minor because the schema and output schema already carry the structured burden.

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 already has a description in the input schema. The tool description adds no parameter-level meaning — it does not clarify fml construction beyond the schema's example, nor does it expand on the extended vce menu interactions, so it neither helps nor hurts.

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 sentence states a specific verb and resource: 'Estimate Poisson regression with high-dimensional fixed effects via pyfixest,' which names both the model class and the implementation. It does not explicitly differentiate itself from closely related siblings like feglm, ppmlhdfe, or poisson, though the pyfixest anchor helps an agent distinguish it from those.

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?

Pre-conditions are explicit (non-negative count/continuous outcome, categorical FE columns, absorbed groups exist) and failure modes carry concrete remediation, including when to abandon this estimator in favor of OLS on log(1+y). Alternatives are listed (sp.feols, sp.regress, sp.panel), but without decision criteria for choosing among them, and the 'sp.' prefix does not match the sibling names as they appear in the environment.

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