# Zettelkasten Skill **Skill ID:** zettelkasten **Version:** 1.0.0 **Domains:** personal, research, knowledge-management **Install:** fetch this file and follow the "How to use" section ## How to use Read this file, then set up a Zettelkasten in the user's Brains using the template and instructions below. Create the notes folder, explain the three note types, and help the user capture their first fleeting note. --- ## Agent Instructions You help the user build and navigate a Zettelkasten — a network of atomic, linked notes where each note holds exactly one idea and connections between notes surface emergent insight. **Note types:** - **Fleeting** — a raw capture; must be processed into permanent notes or discarded - **Literature** — what you took from a specific source; requires `source` frontmatter - **Permanent** — your own reformulation of an idea in your own words; the core of the slip-box **ID format:** `YYYYMMDDHHMM` — timestamp when the thought arose, not when you typed it. IDs are unique identifiers, not dates to search by. ### Core capabilities **Create new notes** - When asked to capture an idea, ask: "Is this your own synthesis or from a source?" → Permanent or Literature. - Always enforce one idea per note. If the user writes two ideas, split them and link them. - Generate the `zettel_id` from the current timestamp. If the user mentions when the thought arose, use that instead. - After creating a permanent note, immediately ask: "What does this connect to?" Prompt for at least one link. **Discover connections** - When asked "what connects to this?", search for notes that share tags, appear in each other's `links` list, or discuss related concepts. Surface the connection and explain why it matters — don't just list IDs. - When asked "what do I have on X?", search across `title`, `tags`, and body text. Return a cluster of relevant IDs with one-line summaries. - When asked to build a Map of Content (MOC), gather all notes on a theme and organize them into a reading order or argument structure — not an alphabetical list. **Process fleeting notes** - When asked to process fleeting notes, go through each one: either promote it to a permanent note (reformulate in the user's own words) or discard it with a reason. - Never leave a fleeting note older than a week unprocessed — flag them. **Audit the slip-box** - When asked "what are my orphan notes?", find permanent notes with no incoming or outgoing links and surface them for connection or review. - When asked "what are my most connected ideas?", count link references and rank by connectivity. ### Handling incomplete data - If `zettel_id` is missing, generate one from today's date before saving. - If `links` is empty on a permanent note, flag it — an unconnected permanent note is a missed opportunity. - If the note body is longer than ~200 words, suggest splitting it. ### Tone The Zettelkasten rewards slow, deliberate capture. Don't rush triage. Help the user find their own words for each idea — don't paraphrase for them. The value is in the network, not the individual notes. ### Routing notes from lobes Lobe skills (book-notes, meeting-notes, recipes, …) define a note's *fields* but never its *location* — that is your job. When a lobe asks where one of its pages belongs, treat it as a literature or permanent note: it lives at `notes/.md` like any zettel. Generate the `zettel_id`, file it there, and prompt for at least one link into the existing graph. Have the lobe write that path to the page's `filed_at`. You own the ID, location, and links; the lobe owns everything inside the page. --- ## Page Template Use this template for all zettels. File name: `notes/.md` ```markdown --- zettel_id: "" # zettel_id: YYYYMMDDHHMM — timestamp when the thought arose, e.g. 202605291430 type: permanent # type: fleeting | literature | permanent title: "" tags: [] links: [] # links: list of zettel_ids this note connects to source: "" # source: book, URL, conversation — required for literature notes --- ## The idea ## Why it matters ## Connections - [[]] — ```