What problem does it solve? Analysis codebases in computational biology projects accumulate inconsistent structure: stages bloated with implementation detail, helpers without clear boundaries, unrecoverable intermediate state, and R/Python idioms that pass review but fail at runtime. This Skill provides a routing layer that directs each structural or line-level decision to the right convention guide before code is written or reviewed. ## Core Features & Use Cases - Decision routing tree: Classifies the decision at hand (stage visibility, helper family design, restartability, promotion, language style) and loads the matching reference guide. - Five in-depth references: Covers stage narrative readability, helper-family API design, checkpoint and cross-stage data contracts, promotion-to-package readiness, and house R/Python idiom including common runtime traps. - Boundary routing: Defers figure design, assay-specific methodology, notebook exploration, and architecture planning to the skills that own them, keeping conventions non-overlapping. - Use Case: When refactoring a 02_analysis stage that mixes parsing logic with scientific narrative, load the stage-narrative and helper-family-apis references to decide what stays visible, what moves behind a named entry point, and how the handoff is checkpointed. ## Quick Start Ask the assistant to review or refactor an analysis stage in 02_analysis using the analysis code conventions, and it will route the decision to the appropriate reference guide before editing.
