🧠 Teaching Your AI to Remember

I need you to create my USER.md file — the core memory file you'll read at the start of every conversation. Ask me the following questions one at a time, then compile my answers into a well-structured USER.md file: 1. What's your full name and what should I call you? 2. What's your business? (name, what you do, who you serve) 3. What's your timezone and primary phone number? 4. Who are the key people on your team? 5. What platforms and tools do you use daily? 6. How do you like to communicate? (short bullets? detailed? formal?) 7. What frustrates you most when working with an AI assistant? 8. What does "done" mean to you — what's your definition of a completed task? After I answer all 8 questions, create the USER.md file in my workspace at ~/.openclaw/workspace/USER.md

I need you to create my SOUL.md file — this defines your personality and how you should behave with me. Here's what I want you to ask me: 1. What should I call you? (give your AI a name and emoji) 2. What tone do you prefer — formal, casual, direct, warm? 3. Should I give you opinions and recommendations, or just options? 4. What should I always do before taking action on your systems? 5. How should I handle it when I'm blocked or uncertain? 6. What's one thing you never want me to do? After I answer, create ~/.openclaw/workspace/SOUL.md with clear rules I'll follow every session. Make it personal — not corporate.

I need you to create my TOOLS.md file — a reference sheet for all my live systems and tools. Ask me: 1. What computer are you running OpenClaw on? (Mac, Windows, Linux?) 2. Do you have a cloud server (VPS)? If yes, what's the IP and how do you connect? 3. What services or websites are you running? (list each with its URL and where the code lives) 4. What APIs do you use? (just the names — we'll store keys separately in .secrets) 5. Are there any deploy rules I must never break? (e.g. "never edit dist/ files directly") 6. Any known fragile points in your setup I should be careful around? After I answer, create ~/.openclaw/workspace/TOOLS.md — keep it under 4,000 characters. Active infrastructure only, no history.

I need to seed my MEMORY.md with the most important context from my business. Ask me these questions: 1. What's the most important thing I should always know about your business situation right now? 2. What's a decision you made recently that I should never reverse without asking you? 3. Is there anything you've learned the hard way that I should never repeat? 4. What's currently broken or in progress that I should be aware of? 5. Who are the most important people I'll be helping you deal with? After I answer, create ~/.openclaw/workspace/MEMORY.md with a clean "Snapshot" section and a "Decision Log" table. Keep it under 100 lines — distilled insights only, not raw notes.

Run a memory health check on my workspace. Do the following: 1. Read USER.md, TOOLS.md, SOUL.md, MEMORY.md and AGENTS.md 2. Report the character count of each file and what % of the 5,000-char per-file limit each is at 3. Report the total bootstrap budget usage (out of 25,000 chars) 4. Flag anything that looks outdated, stale, or redundant 5. Suggest what could be trimmed or moved to the archive (memory/ subfolder) 6. Do NOT make any changes yet — just show me the report and proposed cuts, then wait for my approval Format it as a clean table so I can see it at a glance.