# CLAUDE.md

This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.

## Vault Architecture

This is an Obsidian vault organized using the PARA method:

- **000-inbox/**: Temporary storage for new notes and clippings awaiting processing
- **100-project/**: Active projects with defined outcomes and deadlines
  - `Personal/`: Personal projects (hardware, VPS, cooking, etc.)
  - `Work/`: Work-related projects
  - `Notes Migration/`: Ongoing vault organization work
- **200-area/**: Areas of responsibility and ongoing interests (Blog, Finance, Health, House, Job, etc.)
- **300-resources/**: Reference materials organized by topic (Development, Cooking, Gaming, PKM, etc.)
- **400-archive/**: Completed or inactive items
- **Clippings/**: Web clippings and saved articles
- **ReadItLater Inbox/**: Articles queued for reading

The vault uses Smart Connections (`.smart-env/`) for semantic search with local embeddings (TaylorAI/bge-micro-v2).

## Vault Refactor Operating Rules

### Non-negotiables
- Never delete information. If replacing/merging notes, move originals to `_archive/` and leave a short stub note linking to the merged result.
- Changes must be incremental: one batch = one clear objective + small diff.
- Before any edits: propose a plan and list exact files to touch.
- After edits: produce a change report (moved/renamed/merged files) and run link checks if available.

### Scope
- Treat this Obsidian Vault as data, not an app project.
- Do NOT modify `.obsidian/` unless explicitly asked.

### Refactor conventions
- Use consistent frontmatter keys: `title`, `aliases`, `tags`, `created`, `updated`, `source`.
- For merges: keep canonical note in `Notes/` and archive duplicates in `_archive/duplicates/`.
- For renames/moves: update wikilinks `[[...]]` across the vault.

### Workflow
1) Audit → 2) Plan → 3) Apply one batch → 4) Summarize → 5) Wait for review before next batch