Ink & Memory

中文 · English

Write. Reflect. Listen to the voices within.

Ink & Memory is a journaling studio inspired by the inner monologues of Disco Elysium. As you write, a council of distinct personas chimes in—offering perspective, asking questions, or pointing out the absurdity of it all.

This isn't just another notes app. It's a daily writing companion that saves your work, organizes your thoughts, and helps you notice patterns in your own thinking. It speaks both English and Chinese, switching seamlessly as you write.

---

What Makes It Different

A clean writing space. No toolbar clutter. Just a calm surface for your thoughts, with automatic saving so nothing gets lost.

A council of inner voices. As you type, different personas highlight phrases and offer their takes—from the pattern-spotting Mirror to the darkly funny Absurdist. They appear as gentle watercolor highlights in the margins, never interrupting your flow.

Your timeline, visualized. Every session is saved to a calendar. Each day generates a unique image from your writing—a visual diary that grows with you.

Fully customizable. Don't like a voice? Edit it. Want new perspectives? Create your own deck. Share your creations with the community or discover what others have made.

---

The Voices

The voices are organized into three decks. Enable the ones that resonate; disable the rest.

Introspection Deck

For processing emotions and understanding yourself

	Voice	What it does
❤️	Holder	Offers gentle validation and support
👁️	Mirror	Reflects patterns you might not notice
👊	Starter	Breaks paralysis with tiny first steps
🧭	Weaver	Finds hidden threads connecting your thoughts
🎭	Absurdist	Lightens heaviness with dark humor

Scholar Deck

For intellectual and academic perspectives

	Voice	What it does
🧭	Linguist	Analyzes structure, semantics, meaning
👁️	Painter	Focuses on imagery, aesthetics, mood
💡	Physicist	Applies principles of energy and systems
🧠	Computer Scientist	Thinks in algorithms and complexity
❤️	Doctor	Offers health and psychological angles
🧭	Historian	Provides context and historical patterns

Philosophy Deck

For examining life through different lenses

	Voice	What it does
🛡️	Stoic	Emphasizes what you can and can't control
💨	Taoist	Points toward effortless action and flow
🤔	Existentialist	Asks about choice, freedom, meaning
👊	Pragmatist	Focuses on what actually works

Making Your Own

Each voice is just a system prompt, an icon, and a color. Fork any deck to customize it. Create voices that channel your favorite thinker, focus on specific aspects of your life, or just make you laugh. Share them in the community store.

---

Getting Started

You'll Need

• Python 3.11+ with uv
• Node.js 18+

Backend Setup

cd backend
uv venv
source .venv/bin/activate  # Windows: .venv\Scripts\activate
uv pip install -r requirements.txt

Create a models.json file for your LLM:

{
  "models": {
    "gemini-3-flash-preview": {
      "endpoint": "https://openrouter.ai/api/v1",
      "api_key": "your-key-here",
      "model": "google/gemini-3-flash-preview"
    }
  },
  "roles": {
    "voice_analysis": "gemini-3-flash-preview",
    "voice_chat": "gemini-3-flash-preview"
  }
}

See backend/models.json.example for the full set of roles and image settings.

Then start the server:

python server.py  # Runs on http://localhost:8765

Manual database initialization

You need to create the SQLite tables before first launch:

cd backend
python database.py

Frontend Setup

cd frontend
npm install
npm run dev  # Runs on http://localhost:5173

---

How It Works

The voice system uses a trace-based energy model:

1. Energy builds as you write
2. Threshold triggers analysis of recent text
3. Voices scan for phrases that match their personality
4. Comments appear as watercolor highlights in the margin

This creates an organic rhythm—voices chime in naturally rather than constantly interrupting.

---

Technical Stack

Frontend: React 19 + TypeScript, Vite, TipTap editor with custom extensions

Backend: FastAPI + Python, PolyCLI for LLM orchestration, SQLite with WAL mode

AI: Multi-model support (GPT-4, Claude, DeepSeek, Gemini), structured outputs via Pydantic

Architecture: Deck-based voice system with parent-child relationships, community store for sharing

---

Roadmap

• More voices and community-created decks
• Richer visualizations of your writing patterns
• Mobile app for writing on the go
• Collaborative features

---

Contributing

This is open source. We'd love your help.

• Found a bug? Open an issue
• Have an idea? Let's discuss it
• Want to code? PRs welcome
• Created a cool deck? Share it with the community

---

Your thoughts deserve to be heard—even by yourself.