What problem does it solve? Turning long books, documentation sets, or scattered source material into an operational AI skill is slow and error-prone: summaries lose the actionable rules, examples get dropped, and the result rarely tells an agent what to do or when. This Skill converts source material into a structured, invocation-ready skill package with a concise SKILL.md, dense references, and adapted examples. ## Core Features & Use Cases - Source-to-Skill Distillation: Accepts local files, folders, and URLs, then extracts transferable rules, workflows, checklists, pitfalls, and examples into a complete skill package under .cursor/skills or a custom --path output root. - Split-Skill Mode: Use --split or --split-by <strategy> (auto, goal, topic, workflow, audience) to generate multiple focused child skills sharing one namespace prefix, each with a distinct activation trigger. - Large Material Handling: A bundled material-prep.py script extracts and chunks PDFs, folders, and URLs into a manifest and source index, with safeguards that skip hidden and credential-like files. - Update & Redaction Modes: --update merges new material into an existing skill without duplicating files, and --redact-source-map omits source titles, URLs, and paths from generated output. - Use Case: Point it at a programming book PDF with --name code-quality --split to produce a toolkit of focused review and refactoring skills, each with its own references and before/after code examples. ## Quick Start Ask the agent to distill a book or docs folder into a skill, for example: run aif-distillation on ./books/domain-driven-design.pdf with the name ddd-practices.
