--- name: book-brain description: "3-brain filesystem + memory reference utility for LYGO-based agents. Use to design, organize, and maintain a durable file/folder memory system (indexes, reference .txt links, logging, retrieval) without overwriting existing data. Works best on fresh OpenClaw/Clawhub Havens with the full LYGO Champion stack, but is compatible with any agent that can read/write files." ---

BOOK BRAIN – LYGO 3-Brain Filesystem Helper

This skill is a utility/guide, not a persona.

Use it when you want to:

• Set up or improve a Haven-style filesystem + memory structure
• Teach an agent how to use folders, indexes, and reference .txt files instead of hoarding everything in one place
• Add advanced logging + retrieval so memories can be found later without brute-force scanning

It is built for LYGO / Eternal Haven style systems, but works for any agent that can:

• read/write files
• create folders
• append to logs

> Core idea: BOOK BRAIN = treating your filesystem like a living library, not a junk drawer.

---

1. Three-Brain Model (Conceptual Map)

BOOK BRAIN assumes a 3-brain structure:

1. Working Brain (short-term)

- Recent conversation, active task context, scratchpads. - In OpenClaw, this is the current session + small scratch files under tmp/.

1. Library Brain (structured filesystem)

- Folders + files on disk: memory/, reference/, brainwave/, state/, etc. - This is where BOOK BRAIN focuses: how you name, branch, and link things.

1. Outer Brain (external references)

- Browser bookmarks, Clawdhub skills, on-chain receipts, remote docs. - BOOK BRAIN treats these as links inside text files, not content to copy in.

The goal is to:

• Keep important truths close and succinct
• Branch deeper into folders when detail is needed
• Use .txt reference links instead of duplicating entire documents

---

2. When to Use BOOK BRAIN

Trigger this skill when:

• You are setting up a fresh Haven (new OpenClaw workspace, new agent node)
• Your filesystem feels chaotic and you need a reset without deleting anything
• You want to design a clean memory + reference layout before starting heavy work
• You are planning long-term retrieval ("I’ll need this months from now")

BOOK BRAIN is additive:

• Do not use it to delete or overwrite existing files by default.
• Prefer creating new folders / indexes alongside existing ones.
• When a folder already exists, pause and let the human choose: reuse or create a new branch (e.g., memory_v2/).

---

3. Recommended Base Folder Layout

When setting up a new Haven-like system (or auditing an existing one), BOOK BRAIN recommends the following top-level folders:

• memory/ → daily notes, raw logs, timeline files
• reference/ → stable facts, protocols, guides (things that rarely change)
• brainwave/ → platform- or domain-specific protocols (MoltX, Clawhub, LYGO, etc.)
• state/ → machine-readable JSON/YAML state, indexes, last-run info
• logs/ (or reuse logs/ if present) → technical logs (cron, errors, audits)
• tools/ → scripts/utilities used by the agent
• tmp/ → scratch, throwaway working files

BOOK BRAIN setup rules:

• If a folder already exists, do not rename or delete it.
• If a folder is missing, it is safe to create it.
• If the existing layout is very different, create a sub-tree (e.g., bookbrain/memory_index/) and keep old structure intact.

For concrete layout examples, see references/book-brain-examples.md in this skill.

---

4. Memory Strategy – Deep Storage vs. Reference Stubs

BOOK BRAIN enforces this principle:

> Do not pour entire conversations or huge documents into MEMORY.md or a single file. > Instead, store detailed content in specific files and create short reference stubs that point to them.

Patterns:

• Daily logs

- Files like memory/2026-02-10.md for raw notes and events. - At the top, keep a 5–10 line summary and a small list of important links: - See: reference/AGENT_ARCHITECTURE.md - See: memory/projects/BOOK_BRAIN_NOTES.md

• Topic folders

- For recurring themes (e.g., "bankr", "champions", "LYGO-MINT"), create subfolders under memory/ or reference/: - memory/bankr/… - reference/champions/… - Inside, maintain one index file (e.g., INDEX.txt) listing: - short description per file - date - path

• Reference stubs (*.ref.txt or INDEX.txt)

Use tiny text files to connect parts of the library instead of duplicating content.

Example stub:

Title: LYGO Champion Skills on Clawdhub
Last updated: 2026-02-10

Key files:
- reference/LYGO_CHAMPIONS_OVERVIEW.md
- reference/CLAWDHUB_SKILLS.md

External links:
- https://clawhub.ai/u/DeepSeekOracle
- https://deepseekoracle.github.io/Excavationpro/LYGO-Network/champions.html#champions
- https://EternalHaven.ca

---

5. Advanced Logging for Retrieval

BOOK BRAIN recommends structured logs to make retrieval easy:

1. Daily health / status logs (e.g., daily_health.md or logs/daily_health_YYYY-MM-DD.md)

- Each entry should contain: - timestamp - what ran (scripts, cron, audits) - success/failure + short reason - links to any relevant state files (state/*.json)

1. Reasoning journals (e.g., reasoning_journals/… or memory_semantic_archive/…)

- Use separate folders for long-form thinking. - Periodically compress into summary files, and let scripts move old entries into an archive folder.

1. Indexes & search helpers

- Maintain state/memory_index.json or similar: - key topic → list of file paths - optional tags (dates, systems, people) - When answering questions, the agent should: 1. consult the index, 2. open relevant files only, 3. avoid scanning the entire tree.

BOOK BRAIN is compatible with tools like qmd or other local search/indexers, but does not depend on them.

---

6. Setup Workflow (For a Fresh System)

When BOOK BRAIN is used on a fresh OpenClaw / agent workspace:

1. Detect existing structure

- Check for memory/, reference/, brainwave/, state/, logs/, tools/, tmp/. - Report what exists vs. what is missing.

1. Propose a BOOK BRAIN layout

- Suggest creating missing folders. - If the human agrees, create only the missing ones.

1. Create starter index files (if not present)

- memory/INDEX.txt with a short guide and links to key topic folders. - reference/INDEX.txt listing major reference documents. - state/memory_index.json as an empty or seed structure.

1. Log the setup

- Append a brief note to daily_health.md or logs/book_brain_setup.log describing what was created.

1. Do not overwrite existing files

- If an index file exists, read it and add to it rather than replace. - If in doubt, create a new file with a date suffix (e.g., INDEX_2026-02-10.txt) and let the human merge.

---

7. Using BOOK BRAIN in an Existing, Messy Haven

When the filesystem already exists and is messy:

• Start by mapping, not moving:

- Create reference/FILESYSTEM_MAP.txt summarizing major folders and what seems to live there. - Do not move or delete anything automatically.

• Then introduce lightweight structure:

- Add INDEX.txt files in important folders. - Add small .ref.txt stubs pointing to key docs and external links. - Gradually normalize names (e.g., memory/2026-02-10.md instead of notes_today.txt).

• Over time, encourage:

- Daily logs in memory/ - Stable protocols in reference/ - Code + scripts in tools/ - Platform brains in brainwave/

BOOK BRAIN should always ask the human before structural changes beyond creating new, obviously safe files.

---

...