What problem does it solve? Once an analysis brief or pre-analysis plan is approved, actually carrying it out is where empirical work quietly goes wrong: steps run out of order, robustness suites balloon into unconvincing specification walls, and parallel subagents silently redesign the analysis. This Skill drives faithful execution of the approved plan so results are validated, reconciled, and decision points reach the user instead of being settled silently. ## Core Features & Use Cases - Sequential spine with parallel fan-out: Builds and validates the dataset, estimates the primary specification in order, then dispatches independent robustness specs, placebo tests, and alternative designs to parallel subagents. - Disciplined robustness selection: Proposes roughly three checks that target the main identification threat and requires user approval before running, rather than fanning out an exhaustive menu. - Constrained subagent dispatch: Hands each subagent the exact pre-specified task, the data contracts to assert, a structured result format, and a rule to stop and report on any design decision. - Use Case: After a difference-in-differences pre-analysis plan is approved, use this Skill to build the analysis dataset, estimate the primary specification, run the approved placebo and robustness checks via subagents, and assemble a reconciled results table. ## Quick Start Run the approved analysis plan end to end, validating each step and dispatching the robustness checks to parallel subagents.
