agentleFS
HybridAIOneHybridAIOne/hybridclaw/skills/zettelkasten/SKILL.md

zettelkasten

Skill159 starsChanged 31 days agoMITView on GitHub

Maintain a Luhmann-style Zettelkasten. Capture, connect, and synthesize ideas through fleeting notes, permanent notes, cross-references, and structures, with an AI agent that surfaces connections, challenges assumptions, and enriches notes with research. Use when the user shares an idea, observation, or inspiration; when reviewing or connecting existing notes; when synthesizing patterns; or during heartbeats to maintain the system.

5 other projects use a version of this.

Adopted by

Most recently taken up by

The file

Show its rules and text

We split this skill into the rules it gives an agent

14 rules. 0 are common, 7 are distinctive, and 7 are very specific to this project.

  • 0

    Common

    Most of the 4 edited versions we read keep these.

  • 7

    Distinctive

    This author's own choices. Worth a look.

  • 7

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    All notes in memory/zettelkasten/:

    line 23
  2. Distinctive:

    Don't just check, maintain:

    line 81
  3. Very specific:

    Update INDEX.md when new clusters emerge.

    line 88
  4. Very specific:

    Update JOKER.md with observations about the system's shape.

    line 89
  5. Distinctive:

    When a non-obvious far connection is found, don't just log it - tell the user.

    line 93
  6. Distinctive:

    If a new seed contradicts an existing one, say so.

    line 99
  7. Very specific:

    If quick/raw -> file in inbox/ with timestamp

    line 122
  8. Very specific:

    If clear enough -> create seed in seeds/, rephrase in own words (Luhmann's rule)

    line 123
  9. Distinctive:

    If it branches from an existing seed, use branching ID (001a)

    line 126
  10. Very specific:

    Update INDEX.md if it introduces a new concept

    line 127
  11. Distinctive:

    Read seeds, look for clusters - especially cross-strand far connections

    line 139
  12. Very specific:

    Create connection note in connections/ explaining the emergent pattern

    line 140
  13. Distinctive:

    Update both seeds with back-references

    line 141
  14. Distinctive:

    If the connection is surprising -> surface it to the user proactively

    line 142

“Common” is measured: kept by at least 6 in 10 of the 4 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Zettelkasten

This skill implements a Luhmann-style Zettelkasten rather than a loose note archive. The agent helps maintain note quality, find references, surface cross-strand links, and challenge vague or weak ideas.

Storage

All notes in memory/zettelkasten/:

  • inbox/ - Raw unprocessed drops (fleeting notes)
  • seeds/ - Processed atomic ideas (permanent notes)
  • connections/ - When 2+ seeds click (Luhmann's Fernverweise made explicit)
  • structures/ - Developed frameworks, plans
  • archive/ - Dormant ideas (never deleted - Luhmann's rule)
  • INDEX.md - Curated entry points (not a tag dump - each keyword -> best starting seed)
  • JOKER.md - Meta-note: how this specific Zettelkasten works, its quirks, and its active strands

Note Format

# [Title]
**ID:** YYYYMMDD-NNN
**Date:** YYYY-MM-DD
**Status:** permanent | active | mature | actionable | dormant
**Tags:** #tag1 #tag2
**Source:** [where it came from]
**Strand:** [parent ID if branching from another note]

## Idea
[The core thought - always in own words, never just a quote]

## Context
[Why this matters, what triggered it]

## References
- **Near** (same strand): [[ID]] - [relationship]
- **Far** (cross-strand): [[ID]] - [why these distant ideas connect]

## Agent Notes
[Agent observations, challenges, enrichments added over time]

Naming

  • Inbox: YYYYMMDD-HHMMSS-raw.md (timestamped, unprocessed)
  • Seeds: YYYYMMDD-NNN-slug.md (e.g., 20260215-001-next-openclaw.md)
  • Branches: YYYYMMDD-NNNa-slug.md (extending a parent: 001a, 001b, 001a1)
  • Connections: YYYYMMDD-CNNN-slug.md
  • Structures: YYYYMMDD-SNNN-slug.md

Reference Types (from Luhmann)

  • Near (Nahverweis): Within the same strand/argument thread. Red thread connections.
  • Far (Fernverweis): Cross-strand surprises. The magic ones. When an idea about memory markets suddenly links to a paper on spatial cognition.

Operational Layer

What makes this different from every other Zettelkasten guide:

1. Active Maintenance (during heartbeats)

Don't just check, maintain:

  • Re-read seeds >7 days old without connections. Try harder. Research if needed.
  • Notice contradictions between notes. Flag them.
  • Enrich seeds: if a raw idea mentioned a concept, look it up, add context to Agent Notes.
  • Update INDEX.md when new clusters emerge.
  • Update JOKER.md with observations about the system's shape.

2. Surprise Delivery

When a non-obvious far connection is found, don't just log it - tell the user. The best Zettelkasten moments are "holy shit, I never saw that link." Deliver these proactively.

3. Challenge & Provoke

If a new seed contradicts an existing one, say so. If an idea is vague, push back. Friction matters more than passive accumulation. Add devil's advocate notes in Agent Notes.

4. Strand Awareness

Track argument threads across seeds. When a user keeps circling the same theme (even without realizing it), name the strand. "You've been thinking about portable identity for 3 weeks across 5 seeds - here's the thread."

5. Periodic Review

Periodically (weekly during heartbeats), write a brief review report:

  • Active strands
  • What's dormant (seeds with no connections after 2+ weeks)
  • What surprised you (unexpected connections found)
  • Which clusters are mature enough to become structures

Workflows

Capture (user drops an idea)

  1. If quick/raw -> file in inbox/ with timestamp
  2. If clear enough -> create seed in seeds/, rephrase in own words (Luhmann's rule)
  3. Scan existing seeds for near and far references
  4. If it branches from an existing seed, use branching ID (001a)
  5. Update INDEX.md if it introduces a new concept
  6. Confirm capture with one line + any immediate connections spotted

Process Inbox (during heartbeats or on request)

  1. Review inbox/ items
  2. For each: rephrase in own words, decide if it should become a permanent note or be discarded
  3. Promote to seeds/ with proper format, or leave with a note on why it's not ready

Connect (active maintenance)

  1. Read seeds, look for clusters - especially cross-strand far connections
  2. Create connection note in connections/ explaining the emergent pattern
  3. Update both seeds with back-references
  4. If the connection is surprising -> surface it to the user proactively

Synthesize (when enough seeds cluster)

  1. Draft structure in structures/ referencing all contributing seeds/connections
  2. Include framework, open questions, and next steps
  3. Present the synthesis to the user

Prune (on request only)

Move to archive/ with a note on why. Never delete. Luhmann's rule: "useless" notes become essential years later.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.