{"slug": "the-vault-skeleton-a-second-brain-your-ai-can-actually-read-jlee-mov", "title": "The Vault Skeleton — a second brain your AI can actually read · @jlee.mov", "summary": "A developer known as @jlee.mov shared a folder structure for a 'second brain' vault designed to let AI assistants like Claude navigate a business's notes by lifecycle stage. The system uses numbered folders from 00-Inbox to 07-Analytics, a CLAUDE.md map file, and a memory folder of single-fact files to keep context lean and answers grounded.", "body_md": "The folder structure I use so Claude can navigate my whole business and pull the exact note in one query. Copy it, fill it, keep it fed.\n\n*From @jlee.mov · pairs with CAROUSEL-005.*\n\nEverything is one pipeline. A note moves `00 → 07`\n\nas it goes from raw thought to shipped. **Location tells you (and the AI) the status** — no status field to maintain, no \"misc\" folder to guess through.\n\n```\nmy-vault/\n├── CLAUDE.md            ← the map (see §2)\n├── 00-Inbox/            raw drops, unsorted — clear it weekly\n├── 01-Ideas/            captured + shaped, not started\n├── 02-Drafts/           in progress\n├── 03-Published/        done / shipped\n├── 04-Sources/          raw material, assets, references\n├── 05-Competitors/      research on other people\n├── 06-Hooks/            reusable openers / patterns\n├── 07-Analytics/        what actually worked\n└── _meta/               ops docs — the owner files (see §2)\n```\n\nRename the folders to your work (a consultant might use `03-Delivered`\n\n, `04-Client-Files`\n\n), but keep two rules:\n\n**Numbered by stage**— the number is the lifecycle position.** Same folders every project**— boring and predictable beats clever. The AI learns the shape once.\n\n**Why it works:** a note in `02-Drafts`\n\nis in progress; in `03-Published`\n\n, it's done. You never tell the AI a note's status — it reads it off the path. The structure carries the state so the notes don't have to.\n\n`CLAUDE.md`\n\nsits at the top and is a **map, not a dump**. It doesn't hold your content — it points to the one doc that owns each question. The AI reads the map first, then opens only the file it needs. Context stays lean; answers stay grounded.\n\nGive every recurring question exactly one owner file in `_meta/`\n\n:\n\n| Question | Owner file |\n|---|---|\n| What's next? | `_meta/TODO.md` |\n| Where did I leave off? | `_meta/STATE.md` |\n| What's the strategy / who am I? | `_meta/PROJECT-STATE.md` |\n\n**The rule that keeps it clean:** *one owner per question.* If two files disagree, the owner wins — fix the other on the spot. Two files never quietly contradict each other, so the AI never gets a split answer.\n\nMinimal `CLAUDE.md`\n\nto start:\n\n```\n# [Your business] — brief for the AI\n\nThis is a map, not the content. Read the owner file, not everything.\n\n## Who\n[one line: who you are, what you sell, who for]\n\n## Where things live\n- What's next → _meta/TODO.md\n- Where we left off → _meta/STATE.md\n- Strategy / positioning → _meta/PROJECT-STATE.md\n- Folders are numbered by stage: 00-Inbox … 07-Analytics\n\n## Standing rules\n- [anything the AI should never get wrong: voice, hard no's, who to never email, etc.]\n```\n\nA small `memory/`\n\nfolder of **one-fact files**, indexed by a `MEMORY.md`\n\nlist. These are the things worth never re-explaining — your pricing rule, a client's quirk, a decision you already made.\n\n```\nmemory/\n├── MEMORY.md                    ← index: one line per fact, links to the file\n├── pricing-rules.md             ← one fact\n├── client-acme-prefers-email.md ← one fact\n└── voice-no-emoji-in-body.md    ← one fact\n```\n\nOne fact per file keeps them easy to add, correct, and delete. `MEMORY.md`\n\nis the index the AI scans to decide what's relevant.\n\nThe structure gets the AI *to the right folder*. Semantic search gets it *to the exact note* — by what the note means, not the words you happened to use.\n\n**Easiest:** point**Claude Code**(or Claude Desktop with a folder connected) at the vault. It reads and navigates the files directly.** Add real semantic search**when the vault gets big: a local embeddings-based note search (e.g.`qmd`\n\n, or an Obsidian semantic-search / smart-connections plugin) indexes the folder so a question like*\"how did I price the last client?\"*returns the one right note ranked at the top — instead of 40 keyword matches.\n\nAsk a question → it pulls the exact passage → it answers from your real history. That's the difference between a note app and a second brain.\n\nNone of this works if you stop feeding it. The setup is an afternoon; **keeping notes in their stage is the actual work.** A dead vault is just a prettier version of forgetting. Ten minutes at the end of each day — move what shipped, clear the inbox — is what makes it compound.\n\n*Built this? Reply to the DM and tell me what folder names you landed on — I read every one.*", "url": "https://wpnews.pro/news/the-vault-skeleton-a-second-brain-your-ai-can-actually-read-jlee-mov", "canonical_source": "https://gist.github.com/jleesubai-sys/3d9c3edff184a3084802775c3d65e74a", "published_at": "2026-07-04 04:28:27+00:00", "updated_at": "2026-07-27 22:58:34.408402+00:00", "lang": "en", "topics": ["artificial-intelligence", "developer-tools"], "entities": ["Claude", "jlee.mov", "Claude Code", "Claude Desktop", "Obsidian"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/the-vault-skeleton-a-second-brain-your-ai-can-actually-read-jlee-mov", "markdown": "https://wpnews.pro/news/the-vault-skeleton-a-second-brain-your-ai-can-actually-read-jlee-mov.md", "text": "https://wpnews.pro/news/the-vault-skeleton-a-second-brain-your-ai-can-actually-read-jlee-mov.txt", "jsonld": "https://wpnews.pro/news/the-vault-skeleton-a-second-brain-your-ai-can-actually-read-jlee-mov.jsonld"}}