construct-causal-model

construct-causal-model

Purpose

Construct and validate an explicit causal model with measurable variables, mechanism edges, evidence links, feedback loops, interventions, counterevidence, and confidence.

Input contract

required: [variable_records, mechanism_candidates, evidence_records]
optional: [assumptions, prior_findings, evidence_updates]
constraints: [consume named scientific objects; preserve provenance; keep unresolved uncertainty visible]

Execution protocol

Do not perform called SOP operations inline; each loaded SOP owns its contract and thresholds.

1. You MUST load skill identify-variables to inventory outcomes, factors, mediators, moderators, confounders, and assumptions before drawing any edge.
2. You MUST load skill extract-causal-structure to extract directed cause-mediator-effect chains with temporal order, boundary conditions, and evidence anchors.
3. You MUST load skill represent-mechanism-edge to encode each mechanism edge with pathway, sign, enabling conditions, falsifier, and strength.
4. You MUST load skill attach-evidence-to-relation to attach independent supporting and contradicting evidence while retaining alternative interpretations.
5. You MUST load skill detect-contradiction to compare opposing causal claims under common scope before validation.
6. You MUST load skill detect-feedback-loop to find reinforcing and balancing cycles, delays, uncertain edges, and testable loop implications.
7. You MUST load skill trace-causal-chain to trace target outcomes through intermediate nodes, branches, and because-links.
8. You MUST load skill analyze-intervention to map intervention components, dose, timing, implementation fidelity, comparators, and heterogeneous effects.
9. You MUST load skill construct-counterfactual to propagate a declared intervention and classify the counterfactual outcome.
10. You MUST load skill validate-causal-link to apply CLR-style checks for existence, connection, sufficiency, omitted conditions, and alternatives.
11. You MUST load skill update-confidence-from-evidence to reweight confidence while preserving unresolved conflicts and permitted bounds. If the model should generate testable explanations, consider formulate-hypotheses. If a specific intervention needs minimal-flip and necessity analysis, consider counterfactual-causal-analysis.