What problem does it solve? Analytical work often loses its context: decisions, conventions, and validation steps are forgotten between sessions, leading to irreproducible results and silent statistical errors. This Skill structures every analysis around registered manifests, installed conventions, and mandatory linting so results stay reproducible and auditable. ## Core Features & Use Cases - Convention-routed analysis: Reads project manifests and installed convention packs (robust-analysis, bioinformatics, image-analysis) to apply the correct methodology for clustering, PCA/UMAP, differential expression, survival models, and statistical tests. - Reproducible execution standards: Enforces marimo for exploration, plain Python scripts for pipelines, and a run.sh or run.py entry point that regenerates final outputs. - Static analysis linting: Runs scilintr after every code change to flag silent scientific commitments such as unseeded RNGs, unchecked joins, and magic thresholds, with a structured waiver mechanism. - Use Case: A researcher continuing an RNA-seq differential expression analysis opens the project, and the Skill loads prior decisions from the manifest, applies the installed bioinformatics conventions, registers reportable values, and lints the new code before completion. ## Quick Start Use the analyze skill to run a differential expression analysis on the registered RNA-seq dataset following this project's installed conventions.
