What problem does it solve? Computational chemistry projects often fail because plans, parameters, and conclusions lose traceability between the research question, the actual calculation inputs, and the final manuscript. This Skill turns research goals, structures, software constraints, and real calculation records into a reviewable, staged workflow with a complete provenance chain from scientific question to paper-ready methods section. ## Core Features & Use Cases - Progressive stage-1 intake: Confirms the calculation goal, structure files, software/version, run environment, deliverables, and key parameters before generating any files, then produces a single fixed .review/review.html for final browser approval. - Minimal calculation delivery: After approval, generates only the target program's required input/structure/parameter files plus one README with run commands and return-file instructions; scheduler scripts are created only on request. - Result validation and analysis: Checks returned inputs, outputs, logs, and final structures for completeness and comparability, provides numbered data-processing tutorials, and classifies conclusions as strongly supported, supported, partially supported, inconclusive, or contradicted. - Use Case: A researcher asks to optimize a structure with CP2K. The Skill asks targeted questions until the plan is complete, renders one review page, generates the minimal CP2K input set after approval, then later validates the returned outputs and drafts the manuscript methods section from the real run records. ## Quick Start Load the repository's SKILL.md into your AI tool and say: help me optimize this structure with CP2K, then answer the staged intake questions.
