Skip to main content
Glama
vikranthviki

Causal Decision Agent

by vikranthviki

auto_cate

Read-only

Race multiple meta-learners to estimate heterogeneous treatment effects and return a validated, scored leaderboard with the winning model.

Instructions

Race several meta-learners and return a scored leaderboard + winner. Validation: validated evidence tier (known-truth, reference, external-parity, or Monte Carlo artifact). Assumptions: Unconfoundedness given the covariates; Overlap / positivity across the covariate space; Nuisance functions are estimated consistently; cross-fitting controls overfitting bias. Pre-conditions: Covariates, a treatment indicator, and an outcome for each unit; Enough data to fit flexible nuisance models with sample-splitting / cross-fitting. Failure modes: CATE estimates are unstable or extrapolate beyond the covariate support -> Restrict to the overlap region, increase data, or use a doubly-robust learner (DR-/R-learner). Alternatives: sp.dml, sp.causal_forest, sp.tmle. Typical minimum N: 500.

Input Schema

TableJSON Schema
NameRequiredDescriptionDefault
yYesOutcome column name.
alphaNoSignificance level for both learner confidence intervals and the BLP-beta1 acceptance region used by the selection rule.
scoreNoCurrently only ``'r_loss'`` is implemented. Reserved for future expansion.r_loss
treatYesBinary treatment column name (values in {0, 1}).
detailNoPayload depth: 'minimal' (~150 tokens) for sub-step calls where only the point estimate is needed; 'standard' (~1K tokens) for diagnostics + coefficient table; 'agent' (~2K tokens, default) adds violations / next_steps / suggested_functions so the LLM can plan its next call without another round-trip.agent
n_foldsNoNumber of folds used for both the shared nuisance cross-fit and each learner's honest CATE prediction.
learnersNoShort codes of the meta-learners to race. Duplicates are ignored.
as_handleNoIf true, cache the fitted result on the server and return result_id + result_uri alongside the JSON payload so a subsequent tools/call can chain without re-running.
data_pathYesAbsolute path or URL to a data file. Supported: .csv / .tsv / .txt (delimited), .parquet / .pq, .feather / .arrow, .xlsx / .xls, .dta (Stata), .json / .jsonl. Schemes: file://, s3://, gs://, https://.
result_idNoOptional handle to a previously-fitted result (returned by an earlier call when as_handle=true). Tools that operate on a fitted object accept this in place of re-supplying data_path + columns.
cate_modelNoOverride the default gradient-boosting models used for nuisance and final CATE fitting.
covariatesYesEffect-modifier columns used as features for every nuisance and CATE model.
n_bootstrapNoBootstrap iterations for ATE standard error on non-DR learners (passed through to ``metalearner``).
data_columnsNoOptional column projection. Parquet/Feather/Stata loaders honour this for fast partial reads.
random_stateNoSeed for all K-fold splits.
data_sample_nNoOptional uniform random subsample size (seed=0, deterministic) — useful on huge panels.
outcome_modelNoOverride the default gradient-boosting models used for nuisance and final CATE fitting.
propensity_modelNoOverride the default gradient-boosting models used for nuisance and final CATE fitting.

Output Schema

TableJSON Schema
NameRequiredDescriptionDefault

No arguments

Schema Changelog

Changes observed during successful MCP inspections.

  1. First observedv0.1.0

TDQS

A4.3/5.0
Behavior4/5

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

With readOnlyHint=true and openWorldHint=false already provided, the description adds meaningful behavioral context: assumptions, preconditions, failure modes, and the leaderboard+winner output. It does not contradict the annotations, and the extra context about unstable CATE estimates and overlap support is genuinely useful.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness5/5

Is the description appropriately sized, front-loaded, and free of redundancy?

The description is well structured with labeled sections: purpose, validation, assumptions, preconditions, failure modes, alternatives, and minimum N. It is information-dense but every section earns its place, and the core purpose is front-loaded.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness5/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

Given the 18-parameter complexity and the presence of an output schema, the description covers selection-relevant and invocation-relevant context: when it applies, what data it needs, what can go wrong, and how to react. Nothing essential for an agent to call it correctly is missing.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters3/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema coverage is 100%, so the schema already documents all 18 parameters thoroughly. The description adds only indirect parameter context by naming covariates, treatment, and outcome as preconditions. That meets the baseline but does not go beyond what the schema provides.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

The opening sentence states a specific verb ('Race'), resource ('meta-learners'), and deliverable ('scored leaderboard + winner'). This clearly distinguishes auto_cate from single-learner tools like metalearner or dml. The rest of the description reinforces the scope without blurring it.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines4/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

The description gives clear context through assumptions, pre-conditions, failure modes, and a typical minimum N. It names alternatives (sp.dml, sp.causal_forest, sp.tmle), though it does not spell out exactly when to choose one over this tool or mention closely related siblings like auto_cate_tuned.

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

Deploy Server

Other Tools