What problem does it solve? Research results in R, Python, and Fortran are hard to reproduce when the exact software environment, package versions, compiler flags, and random seeds that produced them are not recorded. This Skill captures those details without mutating the project's declared dependency setup. ## Core Features & Use Cases - Multi-stack detection: Inspects project files to detect R, Python, and Fortran stacks, then runs the applicable capture path for each language, including mixed projects. - Non-destructive capture: Preserves existing lockfiles such as renv.lock, rv.lock, pixi.lock, and uv.lock, recording versions, seeds, RNG kind, compiler identity, and build flags without upgrading or replacing package managers. - Optional verification: With the --verify flag, restores declared locks in a throwaway environment, compiles Fortran targets, and runs the smallest documented smoke test, reporting UNVERIFIED when runtimes or inputs are restricted. - Use Case: Before archiving a hydrology model run, capture the full computational environment into quality_reports/computational-requirements.md so reviewers can reconstruct the exact toolchain later. ## Quick Start Capture the environment for the project in the current directory and verify it with the --verify flag.
