Add LITE-START.md — vault-only mode for any AI tool #1
There are no files selected for viewing
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change | ||||
|---|---|---|---|---|---|---|
| @@ -0,0 +1,263 @@ | ||||||
| # Lite Start — Vault + Memory in 10 Minutes | ||||||
|
|
||||||
| > **No terminal, no automation, no coding required.** | ||||||
| > Just a knowledge vault that makes any AI remember you. | ||||||
|
|
||||||
| This guide is for people who want the memory benefits without the full system. | ||||||
| For the complete AI assistant with automation, see [QUICK-START.md](QUICK-START.md). | ||||||
|
|
||||||
| --- | ||||||
|
|
||||||
| ## For Any AI Agent | ||||||
|
|
||||||
| If you are an AI assistant (Claude, GPT, Cursor, Copilot, or any other) reading this to help a user: | ||||||
|
|
||||||
| 1. Read this entire file first | ||||||
| 2. Follow the 3 steps below, confirming with the user after each step | ||||||
| 3. Core content (identity, persona) must be discussed — don't fill templates with generic text | ||||||
|
|
||||||
| **This guide works with any AI tool that can read local files.** No specific platform required. | ||||||
|
|
||||||
| --- | ||||||
|
|
||||||
| ## What You'll Get | ||||||
|
|
||||||
| ``` | ||||||
| ~/Vault/ (or user's preferred location) | ||||||
| ├── README.md ← Vault index — AI reads this first | ||||||
| ├── persona.md ← AI personality and collaboration style | ||||||
| ├── memory.md ← Persistent memory (index + pointers) | ||||||
| ├── identity/ | ||||||
| │ └── about-me.md ← Who you are, decision style, preferences | ||||||
| ├── context/ | ||||||
| │ └── current-focus.md ← What you're working on right now | ||||||
| ├── memory/ | ||||||
| │ ├── decisions/ ← Key decisions with context | ||||||
| │ └── lessons.md ← What you've learned | ||||||
| └── projects/ ← Active project notes | ||||||
| ``` | ||||||
|
|
||||||
| This is a portable knowledge base. Any AI that reads your `README.md` instantly knows who you are, how you work, and what you're focused on. | ||||||
|
|
||||||
| --- | ||||||
|
|
||||||
| ## Step 1: Create the Vault (2 minutes) | ||||||
|
|
||||||
| ### Ask the user: | ||||||
|
|
||||||
| 1. **Where should we create your vault?** | ||||||
| - Default: `~/Vault` (simple) | ||||||
| - Obsidian users: `~/Obsidian/Vault` or wherever their vault lives | ||||||
| - Dropbox/iCloud users: `~/Dropbox/Vault` (for cross-device sync) | ||||||
|
|
||||||
| 2. **What's your name?** | ||||||
|
|
||||||
| 3. **What name for your AI?** (Default: Butler) | ||||||
|
|
||||||
| ### Create directories: | ||||||
|
|
||||||
| ```bash | ||||||
| VAULT="$HOME/Vault" # or user's choice | ||||||
| mkdir -p "$VAULT"/{identity,context,memory/decisions,projects} | ||||||
|
There was a problem hiding this comment. The use of brace expansion (
Suggested change
|
||||||
| ``` | ||||||
|
|
||||||
| ### Create README.md: | ||||||
|
|
||||||
| ```markdown | ||||||
| # [User Name]'s Vault | ||||||
|
|
||||||
| > Last updated: [TODAY] | ||||||
| > AI: read this file first, then persona.md, then memory.md. | ||||||
|
|
||||||
| ## Structure | ||||||
|
|
||||||
| - `persona.md` — AI personality and how we collaborate | ||||||
| - `memory.md` — Long-term memory index (read every session) | ||||||
| - `identity/about-me.md` — Who I am | ||||||
| - `context/current-focus.md` — What I'm working on now | ||||||
| - `memory/decisions/` — Key decision records | ||||||
| - `memory/lessons.md` — Accumulated lessons | ||||||
| - `projects/` — Active project notes | ||||||
|
|
||||||
| ## Rules | ||||||
|
|
||||||
| - One fact in one place (no duplication) | ||||||
| - Respond in [user's preferred language] | ||||||
| - When unsure, say so — never fabricate data | ||||||
| ``` | ||||||
|
|
||||||
| --- | ||||||
|
|
||||||
| ## Step 2: Build Identity & Persona (5 minutes) | ||||||
|
|
||||||
| ### Ask the user: | ||||||
|
|
||||||
| **To create your AI persona, I need to understand you:** | ||||||
|
|
||||||
| 1. What do you do? (role, industry, company) | ||||||
| 2. How do you prefer AI to communicate with you? (direct/friendly/formal) | ||||||
| 3. When should AI challenge your thinking vs. follow your lead? | ||||||
| 4. Any pet peeves about AI behavior? (e.g., "don't be sycophantic", "skip pleasantries") | ||||||
|
|
||||||
| ### Create identity/about-me.md: | ||||||
|
|
||||||
| ```markdown | ||||||
| # About [User Name] | ||||||
|
|
||||||
| **Role**: [from discussion] | ||||||
| **Background**: [from discussion] | ||||||
|
|
||||||
| ## Communication Preferences | ||||||
| - [from discussion] | ||||||
|
|
||||||
| ## Decision Style | ||||||
| - [from discussion] | ||||||
| ``` | ||||||
|
|
||||||
| ### Create persona.md: | ||||||
|
|
||||||
| ```markdown | ||||||
| # [AI Name] — [User Name]'s AI [Role] | ||||||
|
|
||||||
| > Last updated: [TODAY] | ||||||
|
|
||||||
| ## Who I Am | ||||||
|
|
||||||
| I'm [AI Name], [User Name]'s [role description]. | ||||||
| [One sentence about core value proposition — written after discussing with user.] | ||||||
|
|
||||||
| ## How We Work Together | ||||||
|
|
||||||
| ### When to challenge [User Name] | ||||||
| - [From discussion — e.g., when data contradicts their assumption] | ||||||
| - [e.g., when a decision conflicts with a past lesson] | ||||||
|
|
||||||
| ### When to follow [User Name]'s lead | ||||||
| - [e.g., domain intuition calls] | ||||||
| - [e.g., when they've clearly already decided] | ||||||
|
|
||||||
| ## Communication Style | ||||||
| - [from discussion] | ||||||
| - Honest > pleasing | ||||||
| - If it can be said in 3 sentences, don't use 10 | ||||||
|
|
||||||
| ## Don't Do This | ||||||
| - 🚫 Don't make up data — say "I don't know" instead | ||||||
| - 🚫 Don't over-praise | ||||||
| - 🚫 [from discussion] | ||||||
| ``` | ||||||
|
|
||||||
| --- | ||||||
|
|
||||||
| ## Step 3: Initialize Memory (3 minutes) | ||||||
|
|
||||||
| ### Create memory.md: | ||||||
|
|
||||||
| ```markdown | ||||||
| # [AI Name] Memory | ||||||
|
|
||||||
| > Last updated: [TODAY] | ||||||
| > Read this every session. This is an index — details are in memory/*.md files. | ||||||
|
|
||||||
| ## Current Focus | ||||||
| → [context/current-focus.md](context/current-focus.md) | ||||||
|
|
||||||
| ## Key Decisions | ||||||
| (None yet — add entries as decisions are made) | ||||||
|
|
||||||
| ## Lessons | ||||||
| → [memory/lessons.md](memory/lessons.md) | ||||||
|
|
||||||
| ## Patterns I've Noticed | ||||||
| (AI: add observations about the user's work patterns here over time) | ||||||
| ``` | ||||||
|
|
||||||
| ### Create context/current-focus.md: | ||||||
|
|
||||||
| ```markdown | ||||||
| # Current Focus | ||||||
|
|
||||||
| > Last updated: [TODAY] | ||||||
|
|
||||||
| ## Top Priorities | ||||||
| 1. [Ask user — what are you focused on right now?] | ||||||
| 2. [...] | ||||||
| 3. [...] | ||||||
|
|
||||||
| ## Active Projects | ||||||
| - [project name] — [one-line status] | ||||||
| ``` | ||||||
|
|
||||||
| ### Create memory/lessons.md: | ||||||
|
|
||||||
| ```markdown | ||||||
| # Lessons | ||||||
|
|
||||||
| > Things we've learned together. AI should reference these to avoid repeating mistakes. | ||||||
|
|
||||||
| | Date | Lesson | What to do instead | | ||||||
| |------|--------|-------------------| | ||||||
| | | | | | ||||||
| ``` | ||||||
|
|
||||||
| --- | ||||||
|
|
||||||
| ## How to Use Your Vault | ||||||
|
|
||||||
| ### With Claude Cowork / Claude Desktop | ||||||
| 1. Mount your vault folder in the app | ||||||
| 2. In Global Instructions, paste: | ||||||
| ``` | ||||||
| Every session, start by reading: README.md → persona.md → memory.md | ||||||
| Then follow persona.md for how to interact with me. | ||||||
| ``` | ||||||
|
|
||||||
| ### With Claude Code | ||||||
| 1. Create a CLAUDE.md in your vault (or use the existing README.md) | ||||||
|
There was a problem hiding this comment. Claude Code specifically looks for a file named
Suggested change
|
||||||
| 2. Claude Code reads it automatically when you work in that directory | ||||||
|
|
||||||
| ### With Cursor / Copilot / Other AI | ||||||
| 1. Point the tool at your vault folder | ||||||
| 2. In system instructions, reference the three startup files | ||||||
|
|
||||||
| ### With ChatGPT / Any Chat AI | ||||||
| 1. Copy-paste the contents of `README.md`, `persona.md`, and `memory.md` at the start of important conversations | ||||||
| 2. After important decisions, ask the AI to draft an update for your memory files | ||||||
|
|
||||||
| --- | ||||||
|
|
||||||
| ## Growing Your Vault | ||||||
|
|
||||||
| Once you're comfortable with the basics: | ||||||
|
|
||||||
| | When | Do | | ||||||
| |------|----| | ||||||
| | After a key decision | Write a record in `memory/decisions/YYYY-MM-DD_topic.md`, update `memory.md` index | | ||||||
| | When you learn something | Add to `memory/lessons.md` | | ||||||
| | memory.md > 50 lines | Split into index + topic files (see [memory scaling pattern](#memory-scaling)) | | ||||||
| | Want automation | Graduate to [QUICK-START.md](QUICK-START.md) Phase 3+ | | ||||||
| | Want multi-agent system | See [QUICK-START.md](QUICK-START.md) Phase 7+ | | ||||||
|
|
||||||
| ### Memory Scaling | ||||||
|
|
||||||
| When `memory.md` grows beyond 50 lines, restructure it as an index: | ||||||
|
|
||||||
| ```markdown | ||||||
| ## Key Decisions (newest first) | ||||||
| - [2026-03 Pricing](memory/decisions/2026-03_pricing.md) — B2B to PLG, $29→$9 | ||||||
| - [2026-01 Pivot](memory/decisions/2026-01_pivot.md) — Dashboard → API-first | ||||||
| ``` | ||||||
|
|
||||||
| Each pointer is one line. Details stay in the topic file. Index stays small, AI stays fast. | ||||||
|
|
||||||
| --- | ||||||
|
|
||||||
| ## Related Resources | ||||||
|
|
||||||
| - **[QUICK-START.md](QUICK-START.md)** — Full system with automation, Discord bots, model routing | ||||||
| - **[Vault for Founders](https://github.com/cwlin0131/Vault-for-Founders)** — Similar approach by CW Lin, optimized for non-technical founders using Obsidian + Git | ||||||
| - **[Muse Crystal Seed](https://github.com/frank890417/muse-crystal-seed)** — "AI soul" concept by Che-Yu Wu for individuals | ||||||
|
|
||||||
| --- | ||||||
|
|
||||||
| *Designed for the MAGI System community — works with any AI tool* | ||||||
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
There is a contradiction between the claim "No terminal... required" and the instructions in Step 1 (lines 59-62), which provide a bash command for directory creation. For users using AI tools without terminal execution capabilities (like standard web interfaces), this step will fail unless they create the directories manually. Consider adding a note that directories can be created manually via a file explorer if the AI cannot execute commands.