jd-suggestlisted

# Suggest a Johnny Decimal Taxonomy Analyze a directory tree and propose a JD structure (areas, categories, IDs) that fits the contents. ## Inputs The user provides a directory path (or defaults to the capture inbox / unsorted folder). ## Process ### 1. Scan the directory Use `find` to get a representative sample of paths: ```bash find <path> -maxdepth 5 -not -path '*/.git/*' -not -path '*/node_modules/*' -not -path '*/.venv/*' -not -path '*/__pycache__/*' -not -path '*/site-packages/*' | shuf | head -200 ``` If the directory is inside an existing JD tree, also run `jd triage` and `jd ls` for context. ### 2. Analyze and categorize From the sampled paths, infer the user's activities, projects, and content types. Key principles: - **Think broadly about areas.** Areas should be very broad life domains (Home, Work, Hobbies, Learning), not specific projects. Max 10. - **Categories are where work happens.** Each category should represent a type of activity or content. Max 10 per area. - **IDs are specific things.** Each ID is a single topic, project, or item. Start numbering at .11 (standard zeros .00-.10 are reserved). - **Underfitting is better than overfitting.** Prefer fewer, broader categories over many narrow ones. You can always split later. - **Sampling bias is real.** A Python venv or node_modules will dominate the sample. Weight by uniqueness of path structure, not raw count. - **Infer purpose from context.** A file inside `receipts/2024/` is financial; a file