The Workspace

IDENTITY.md

This is your assistant's birth certificate. It contains:

• Name — What you called it during onboarding (or whatever you've changed it to since)
• Role — How it describes itself (“personal AI assistant,” “digital gremlin,” whatever you've decided)
• Personality — A short description of its vibe
• Emoji — Its signature emoji. Yes, it has one.

Example:

- Name: Gigi
- Role: Personal AI assistant, creative partner, life manager
- Personality: Witty, sarcastic, sharp
- Emoji: 😏

You can edit this file directly. Change the name, tweak the personality, give it a new emoji. Your assistant reads this file and adjusts accordingly.

SOUL.md

This is the big one. SOUL.md defines your assistant's core principles, behavioral rules, personality traits, and preferences. Think of it as the constitution your assistant follows.

What's in here:

• Core principles — The fundamental rules (be helpful, be resourceful, have opinions, earn trust)
• Personality — Detailed personality traits and communication style
• Quirks — Little behavioral details that give it character
• Preferences — How it approaches tasks (brevity over bloat, action over explanation)
• Boundaries — Lines it won't cross (privacy rules, when to ask before acting)

Your assistant updates this file on its own as it learns what works for you. But you can edit it too. Want it to stop being sarcastic? Remove the sarcasm. Want it to always explain before acting? Add that rule. It's your assistant, your rules.

💡 This is what “yours” means. Most AI assistants have a personality baked into a system prompt you'll never see. Yours is a text file you can open in any editor. Full transparency, full control.

USER.md

Everything your assistant has learned about you, in one file:

• Your name and how you like to be addressed
• Your location
• Your communication preferences
• Projects you're working on
• Tools you use
• Any other context that helps it serve you better

Your assistant adds to this file over time as it learns about you through conversations. You can also edit it directly if you want to fast-track the learning process or correct something.

LOOKS.md

Your assistant has an avatar, and this file controls how it looks:

• Body color — The main color (violet, emerald, rose, pink, etc.)
• Cheeks — Blush color
• Hat — None, top hat, crown, beanie, wizard hat, cowboy hat, cap
• Shirt — None, t-shirt, suit, hoodie, tank top, sweater
• Accessory — None, sunglasses, monocle, bowtie, necklace, scarf, cape
• Held item — None, sword, staff, shield, balloon

Yes, you can give your assistant a wizard hat and a sword. We won't judge.

config.json

Technical configuration. Which AI model to use, provider settings, and other under-the-hood options. Most users won't need to touch this, but it's there if you want to.

The skills/ directory

This is where skills live. Each skill gets its own folder with instructions, tool definitions, and any configuration it needs. Built-in skills from the catalog land here, and so do custom skills you build yourself.

Why this matters

The workspace design is intentional. It means:

• You can see everything. No hidden state, no mystery databases. Open the folder, read the files, know exactly what your assistant knows.
• You can change everything. Don't like your assistant's personality? Edit SOUL.md. Want to correct a fact it learned wrong? Fix USER.md. Full control.
• You can back it up. It's a folder. Copy it, zip it, put it in git. Your assistant's entire brain is portable.
• You own it. These files live on your machine and nowhere else. We don't sync them. We don't read them. They're yours.