What problem does it solve? Research projects accumulate scripts in Python, R, and Fortran with undocumented data flows, making replication packages, handoffs, and pipeline audits painful. This Skill reads the code without running it and produces a pipeline.md manifest that documents exactly how every artifact is produced. ## Core Features & Use Cases - Static Dependency Extraction: Records each script's purpose, entry command, files read and written, invoked programs, compiler flags, and figures or tables fed to Quarto or LaTeX documents, including Fortran I/O patterns like open, read, write, and inquire. - Dependency Graph and Run Order: Matches inputs to producers, builds a directed graph, runs a topological sort, and reports cycles, missing inputs, orphan scripts, and duplicate producers. - Structured Manifest Output: Writes pipeline.md with a pipeline table, figure and table manifest, Mermaid or adjacency-list graph, parallel execution groups, diagnostics, and RO-Crate readiness notes. - Use Case: Before submitting a replication package, run this Skill on the project directory to generate pipeline.md documenting every script's role and a verified execution order, then copy the bundled run-all.sh template once the graph is acyclic. ## Quick Start Analyze the scripts in this project directory and generate a pipeline.md manifest with the dependency graph and execution order.
