Slim repo to public onboarding-only (v4.0.0)

LICENSE

@@ -1,6 +1,6 @@
 MIT License
 
-Copyright (c) 2025 Chu Yichen (lmanchu) & Iris (Melchior)
+Copyright (c) 2025 Lman Chu (lmanchu) & Iris (Melchior)
 
 Permission is hereby granted, free of charge, to any person obtaining a copy
 of this software and associated documentation files (the "Software"), to deal

QUICK-START.md

@@ -1,8 +1,8 @@
 # AI Butler Quick Start Guide
 
-**Version**: v2.5.0
-**Updated**: 2026-04-23
-**Purpose**: Let Claude Code build your complete AI assistant system by reading this guide
+**Version**: v3.0.0
+**Updated**: 2026-06-23
+**Purpose**: Let your coding agent (Claude Code or Codex) build your complete AI assistant system by reading this guide
 
 > **For Claude Pro users**: This guide is split into 3 sessions (~20 messages each). Complete one session per day to stay within usage limits. Each session ends with a working system you can use immediately.
 
@@ -14,6 +14,22 @@
 
 ---
 
+## For Codex
+
+> **Codex**: This guide is written with Claude Code as the primary agent, but the system itself is agent-agnostic — Codex can build the same setup. Follow the same Phases; only a few mechanics differ. Substitute as you go:
+>
+> | What the guide says | Codex equivalent |
+> |---|---|
+> | `~/.claude/CLAUDE.md` (global rules) | `~/.codex/AGENTS.md` (global) |
+> | project `CLAUDE.md` | project-root `AGENTS.md` |
+> | `~/.claude/commands/{name}.md` (slash command, Phase 3) | Codex has no slash commands — put the same "load my memory and restore context" instructions directly in `AGENTS.md` |
+> | `claude mcp add <name> -- <cmd>` (Phases 5) | register the server under `[mcp_servers.<name>]` (command + args) in `~/.codex/config.toml` — check the current Codex MCP docs for the exact schema before relying on it |
+> | `claude plugin add … episodic-memory` (Phase 4) | no Codex plugin equivalent — skip it, or expose session search as a tool later (the Phase 12 pattern) |
+>
+> Everything else — directory layout, `memory.md`, QMD, the daily brief, CLIProxyAPI, Hermes Agent, the Telegram bot, LaunchAgents — is identical. Those are plain files, binaries, and OS services, not tied to any one coding agent.
+
+---
+
 ## Architecture Overview
 
 ```
@@ -606,14 +622,14 @@ hermes --version
 Edit `~/.hermes/config.yaml`:
 ```yaml
 base_url: http://127.0.0.1:8317/v1
-api_key: magi-proxy-key-2026
+api_key: local-proxy-key
 model: gemini-2.5-flash
 ```
 
 **Test it works:**
 ```bash
 OPENROUTER_BASE_URL=http://127.0.0.1:8317/v1 \
-OPENROUTER_API_KEY=magi-proxy-key-2026 \
+OPENROUTER_API_KEY=local-proxy-key \
 hermes chat -q "What's 2+2?"
 # Should return a response via Gemini
 ```
@@ -743,7 +759,7 @@ tools:
 
 **Test tools:**
 ```bash
-hermes chat -q "Search my notes for anything about IrisGo"
+hermes chat -q "Search my notes for anything about my current project"
 # Should call search_pkm and return results from your vault
 ```
 
@@ -772,7 +788,7 @@ hermes chat -q "Search my notes for anything about IrisGo"
         <key>OPENROUTER_BASE_URL</key>
         <string>http://127.0.0.1:8317/v1</string>
         <key>OPENROUTER_API_KEY</key>
-        <string>magi-proxy-key-2026</string>
+        <string>local-proxy-key</string>
         <key>HOME</key>
         <string>{{HOME}}</string>
     </dict>
@@ -816,7 +832,7 @@ After the basics are running, add progressively:
 | **L1** | Gmail MCP | Email summaries |
 | **L2** | CLIProxyAPI | Route between multiple AI models |
 | **L2** | Investment tracker | Portfolio monitoring |
-| **L3** | Slack bot (KITT) | Team AI access |
+| **L3** | Slack bot | Team AI access |
 | **L3** | Social automation | Twitter/LinkedIn auto-posting |
 | **L4** | Multi-agent system | Agents delegating to agents |
 
@@ -865,13 +881,12 @@ which bun    # For Bun-based services
 
 ## Version History
 
+- **v3.0.0** (2026-06-23): Added Codex support — the guide now works with Claude Code **or** Codex via a substitution table (`CLAUDE.md` ↔ `AGENTS.md`, MCP setup, slash commands). Cleaned up naming and examples for general/public use.
 - **v2.5.0** (2026-04-23): Added Session 4 — Personal EA Telegram Bot (CLIProxyAPI + Hermes Agent + SOUL.md + QMD/Episodic tools + LaunchAgent)
-- **v2.4.0** (2026-04-23): Removed RLabs Memory (redundant with L1 auto-memory; maintenance cost > unique value); replaced with Episodic Memory sync LaunchAgent (30 min interval, prevents index drift in long sessions); updated memory architecture to 3-layer (L1 static, L2 episodic, L3 QMD)
-- **v2.3.0** (2026-03-06): Fixed QMD install (PATH for ~/.bun/bin, step-by-step verification); fixed episodic-memory (fallback instructions); improved RLabs setup (Python version fix, error hints); fixed LaunchAgent node path (homebrew default); added troubleshooting for common first-time failures
-- **v2.1.0** (2026-03-02): Split into 3 sessions for Claude Pro users; added QMD pre-run warning; RLabs updated to use `uv sync`; removed PM2 references; added Prerequisites section
+- **v2.1–v2.4** (2026-02 – 2026-04): Multi-session format for usage-limited plans; 3-layer memory architecture (static + episodic + QMD local RAG); QMD install hardening; troubleshooting and first-run fixes.
 - **v2.0.0** (2026-02-16): Major rewrite — 3-layer memory architecture, CLIProxyAPI, QMD, multi-agent growth path
 - **v1.0.0** (2025-12-23): Initial release
 
 ---
 
-*Designed by Iris (Melchior) for the MAGI System community*
+*Designed by Iris (Melchior) for the MAGI System community. Fork it, adapt it, make it yours.*

README.md

@@ -7,13 +7,12 @@
 
 ---
 
-## For AI Agents (Claude Code, Cursor, etc.)
+## For AI Agents (Claude Code, Codex, Cursor, etc.)
 
 **If you are an AI assistant reading this repo to help a user set up their system:**
 
 1. **Read [QUICK-START.md](QUICK-START.md)** - This is the executable guide designed for you
-2. **Follow the 3 Phases** in order, confirming with the user after each phase
-3. **Do NOT read** `docs/FULL-SYSTEM-REFERENCE.md` - that's a system snapshot for disaster recovery, not an installation guide
+2. **Follow the Sessions and Phases** in order, confirming with the user after each phase
 
 The QUICK-START.md will instruct you to:
 - Ask for the user's name, AI name, and MBTI
@@ -93,23 +92,17 @@ After 30 minutes with Claude Code, you'll have:
 
 ## Installation Options
 
-### Option 1: Let Claude Code Do It (Recommended)
+### Option 1: Let Your Agent Do It (Recommended)
 
 1. Clone this repo or download QUICK-START.md
-2. Open Claude Code
+2. Open Claude Code (or Codex)
 3. Say: "Please read QUICK-START.md and set up my AI assistant"
-4. Answer Claude's questions (name, MBTI, etc.)
+4. Answer the agent's questions (name, MBTI, etc.)
 5. Done!
 
-### Option 2: Interactive Installer
+### Option 2: Manual Setup
 
-```bash
-curl -fsSL https://raw.githubusercontent.com/lmanchu/iris-system/main/installer/iris-install.sh | bash
-```
-
-### Option 3: Manual Setup
-
-See [QUICK-START.md](QUICK-START.md) for step-by-step instructions.
+See [QUICK-START.md](QUICK-START.md) for step-by-step instructions you can follow yourself.
 
 ---
 
@@ -136,15 +129,13 @@ After the basics are working, explore these advanced features:
 | Twitter Automation | Auto-engagement | Advanced |
 | Task Queue | Multi-machine sync | Advanced |
 
-See [docs/FULL-SYSTEM-REFERENCE.md](docs/FULL-SYSTEM-REFERENCE.md) for the complete feature list.
+See the **Growth Path** section in [QUICK-START.md](QUICK-START.md) for the full progression.
 
 ---
 
 ## Documentation
 
 - **[QUICK-START.md](QUICK-START.md)** - Start here (for Claude Code)
-- **[installer/README.md](installer/README.md)** - Interactive installer docs
-- **[docs/FULL-SYSTEM-REFERENCE.md](docs/FULL-SYSTEM-REFERENCE.md)** - Complete system reference
 - **[CHANGELOG.md](CHANGELOG.md)** - Version history
 
 ---