Second Brain Template

This repository is a generic template for an Obsidian-compatible second brain. It is designed to preserve messy source material first, then gradually turn it into concise, linked wiki pages that compound over time.

The template is intentionally domain-agnostic. A vault created from it could be for research, school, wedding planning, a company, a creative project, health records, family history, or any other knowledge domain. The categories should change. The transport mechanism should stay stable.

Core Idea

Use the same lifecycle in every vault:

new source material
  -> raw/unprocessed/
  -> reviewed by a person or agent
  -> durable synthesis in wiki/
  -> original preserved in raw/processed/
  -> index and related links updated

The folder names are stable, but the categories inside wiki/ should be customized for each vault.

What Makes It Work

The important reusable system is not the starter taxonomy. It is the way context moves through the vault:

AGENTS.md
  -> the always-loaded contract: source-of-truth pointers, lifecycle, and wiki rules
.agents/skills/
  -> runs repeatable workflows such as ingest, synthesis, and maintenance
.agents/automations/
  -> defines recurring agent work such as daily curation
schema/
  -> stores the durable operating model humans can inspect and revise
raw/
  -> preserves source material
wiki/
  -> accumulates maintained synthesis
wiki/index.md and wiki/log.md
  -> keep navigation and change history alive

This is the compounding loop: every ingest, maintenance pass, and useful answer should leave the wiki slightly easier to use next time.

Structure

AGENTS.md
.agents/
  automations/
  skills/
attachments/
raw/
  unprocessed/
  processed/
schema/
wiki/
  index.md

• AGENTS.md is the always-loaded agent contract. It carries the full rule set.
• .agents/automations/ contains portable automation definitions.
• .agents/skills/ contains repeatable workflows.
• schema/ contains the human-readable source of truth for how the vault works.
• raw/unprocessed/ is the inbox for source material.
• raw/processed/ stores reviewed originals.
• wiki/ contains curated pages.
• attachments/ is for shared supporting files when they are not bundled with a raw source.

Starting A New Vault

1. Copy this template into a new folder.
2. Read schema/adapt-this-template.md.
3. Edit schema/category-map.md for the new vault's domain.
4. Adapt AGENTS.md if any domain-specific rule belongs in the always-loaded contract; keep it short.
5. Add repeatable procedures to .agents/skills/.
6. Configure a daily curation automation from .agents/automations/daily-curation.md.
7. Put new source material into raw/unprocessed/.
8. Ask an agent to use the ingest workflow when you are ready to curate.

Bootstrap Prompt

Use a prompt like this when pointing an agent at a new folder:

Use this second-brain template for this folder. Adapt the category map and wiki
index to the folder's domain, but preserve the raw/unprocessed -> raw/processed
-> wiki lifecycle, AGENTS.md contract, .agents skills, and daily curation
automation pattern. Keep AGENTS.md small. Put domain-specific behavior in
schema, skills, or automation definitions.