A deeply personal, narrative-driven psychological game built entirely in Python, harnessing user data, emotion, and advanced AI-driven NPC interactions.
Maze of Me is a command-line psychological adventure where your digital self is the foundation for a unique, emotional narrative. The game adapts to your profile, calendar, moods, and social connections, creating a different journey every session. Navigate a procedurally generated text maze, meet AI figures who remember your feelings and contacts, and discover how your online presence shapes your subconscious world.
OAuth logins (Spotify, Google, with Facebook & Instagram coming soon) bring in real data to create an interactive and personal experience.
Players awake in a mysterious maze. The story and rooms reference your real-world data, events, moods, and contacts, integrating your digital and emotional life into an interactive psychological narrative:
-
Act 1: Awakening
Bright rooms and supportive AI figures welcome you. Upbeat music, selected from your top Spotify tracks and YouTube history, enhances the mood. -
Act 2: Whispers
Rooms and AI figures reference real calendar events, friends, or memories. NPCs become more challenging, sometimes asking about people in your contacts or your recent mood. -
Act 3: Collapse and Reflection
The maze responds to your feedback, looping through memories and feelings until you reach deeper self-understanding or choose to start over.
- Google OAuth
- Fetches: profile info, calendar events, YouTube watch history, contacts (names, emails, birthdays), Gmail subjects (last 5), Google Tasks (top 10), and recently watched YouTube channels.
- All data is merged into your game profile and used for richer room/NPC generation.
- Spotify OAuth
- Collects: your top tracks, audio features (valence, energy), top artists, favorite genres, liked tracks, and playlists. All data is mapped to in-game moods and events for music and narrative context.
- Procedural Maze & Room Generation
- Rooms are now generated using a much wider set of hooks: your calendar events, YouTube activity, contacts, Gmail, playlists, genres, and more.
- NPCs and room descriptions may reference your favorite artists, email subjects, playlists, and tasks.
- Room hooks and NPC context are more diverse, allowing each playthrough to be completely unique and personal.
- Music as Mood
- Each room's emotion is matched to your music using valence and energy from Spotify data.
- Audio is preloaded using
yt-dlpfor instant playback and played withpygame. - Previous room's audio files are deleted automatically for optimal performance and minimal disk usage.
- Advanced AI NPCs & Memory
- NPCs now have “memory”: they can recall your previous emotions, feedback, and even mention earlier NPC exchanges, moods, or player choices.
- Dialogue is now more deeply personalized—NPCs may reference your recent moods (“I remember you were angry earlier…”), calendar events, contacts, or even subject lines from your recent emails.
- Dialogue choices and emotional feedback influence subsequent AI responses, not just in the current room, but also across your whole session.
- Inspect, Use, and Interact with Room Items
- You can now inspect or “use” furniture in each room for additional AI/NPC context, which may reference your personal data or prior actions.
- Interactions are richer, sometimes involving a “mini-game” or riddle based on your current mood or room context.
- Persistent Save/Load Sessions
- The game now supports saving and continuing your journey later, preserving your room history, moods, and all NPC memory.
- Stats and mood breakdowns are available at any time to track your psychological journey.
- Robust Data & Performance
- Previous room audio files are deleted on transition for disk and performance optimization.
- All major Google and Spotify data fields are now loaded, cached, and referenced in both gameplay and AI logic.
- AI Loading Spinner
- Whenever the AI NPC is generating a response, a live spinner keeps the interface responsive.
- Modern CLI with color, typewriter effect, clear menus, and easy navigation.
- Help menu, graceful exits, and robust error handling.
- Detailed logging of all interactions for transparency and debugging.
- Programming Language: Python 3.13.3
- CLI Engine: Custom-built (input, print, color, spinner, typewriter)
- OAuth Libraries:
spotipy,google-auth,requests-oauthlib - Audio Playback:
pygame,yt-dlp - AI Integration: Local LLM with
llama-cpp-python(Phi-3, LLaMA, Mistral), Ollama, or LM Studio - Data Storage: JSON files
- Utilities: Helpers for parsing, data, and OS/platform detection
- Cross-Platform: Runs on Windows, Linux, and macOS
graph TD
subgraph OAuth["🔐 OAuth Providers"]
Spotify[Spotify OAuth]
Google[Google OAuth]
Facebook[Facebook OAuth planned]
Instagram[Instagram OAuth planned]
end
subgraph Profile["🧍 User Profile"]
Contacts[Google Contacts]
Gmail[Gmail Subjects]
Tasks[Google Tasks]
Playlists[Spotify Playlists]
Genres[Spotify Genres]
TopArtist[Spotify Top Artist]
Liked[Spotify Liked Tracks]
YTChannels[YouTube Channels]
SpotifyData[Spotify Top Tracks + Features]
GoogleCal[Google Calendar Events]
YouTube[YouTube Watch History]
Email[Google Profile Info]
end
subgraph CLI["🖥️ Maze CLI"]
Start[Start Maze]
Move[Choose Direction]
Talk[Talk to NPC]
Inspect[Inspect Room/Furniture]
Log[Interaction Log & Emotion History]
UX[CLI Help, Typewriter FX, Spinner, Exit]
Stats[Stats & Mood Breakdown]
SaveLoad[Save/Load Game State]
MiniGame[Mini-game Events]
end
subgraph LLM["🧠 Local AI"]
PromptGen[Build Prompt: Contacts, Emotions, Events, Tasks, Music]
RoomGen[Room Description]
NPCGen[NPC Dialogue + Memory + Data Hooks]
end
subgraph Audio["🔊 Audio Engine"]
MoodDetect[Analyze Spotify Valence/Energy]
TrackPlay[Play YouTube Audio via yt-dlp]
Cache[Preload & Cleanup Tracks]
end
subgraph Future["🚧 Planned Features"]
FBData[Facebook Data]
IGData[Instagram Data]
NPCFriends[NPCs from Real People]
WebUI[Web-based GUI optional]
end
Spotify --> SpotifyData
Spotify --> Playlists
Spotify --> Genres
Spotify --> TopArtist
Spotify --> Liked
Google --> Email
Google --> GoogleCal
Google --> YouTube
Google --> Contacts
Google --> Gmail
Google --> Tasks
Google --> YTChannels
Facebook --> FBData
Instagram --> IGData
SpotifyData --> Profile
Playlists --> Profile
Genres --> Profile
TopArtist --> Profile
Liked --> Profile
GoogleCal --> Profile
YouTube --> Profile
Email --> Profile
Contacts --> Profile
Gmail --> Profile
Tasks --> Profile
YTChannels --> Profile
FBData --> Profile
IGData --> Profile
Profile --> PromptGen
PromptGen --> RoomGen
PromptGen --> NPCGen
RoomGen --> Start
NPCGen --> Talk
RoomGen --> Inspect
Start --> Move
Move --> RoomGen
Talk --> NPCGen
Inspect --> NPCGen
Log --> NPCGen
Stats --> UX
SaveLoad --> UX
MiniGame --> UX
Profile --> MoodDetect
MoodDetect --> TrackPlay
TrackPlay --> Cache
FBData --> Future
IGData --> Future
Future --> NPCFriends
Future --> WebUI
2025-05-19.02-49-20.mp4
2025-05-20.02-29-50.mp4
git clone https://github.com/bakill3/maze-of-me.git
cd maze_of_me
### 2. Install dependencies:
```bash
pip install -r requirements.txtCreate a .env file in the root directory with:
SPOTIFY_CLIENT_ID=your_spotify_client_id
SPOTIFY_CLIENT_SECRET=your_spotify_secret
GOOGLE_CLIENT_ID=your_google_client_id
GOOGLE_CLIENT_SECRET=your_google_secretTo download the required language model, run download.bat inside the models folder before starting the game.
python cli.py- Spotify and Google OAuth & Data Collection
- YouTube Audio Preloading, Caching, and Cleanup
- Emotion-driven Room and Music Generation
- AI NPCs Powered by Local LLM, with Memory and Contacts
- Dialogue Trees & Player Emotion Feedback
- Loading Spinner for AI Responses
- Inspect & Use Room Items
- Per-Room Audio Cleanup for Performance
- NPCs Reference Contacts, Real Events, and Player Emotions
- Save & load full session, stats, and persistent NPC memory
- Gmail, Google Tasks, and YouTube channel data included in room/NPC logic
- Mini-games and dynamic item interactions
- Facebook & Instagram Integration (planned)
- Persistent Cross-Session NPC Memory (planned)
- Optional Web-based GUI (planned)
Unit tests for all critical parsing and CLI logic.
To run tests:
pytest tests/test_parsers.pyMaze of Me places the highest priority on user privacy and data security:
- All user data collected via OAuth or manual input is stored locally and never shared externally.
- OAuth authentication tokens and sensitive data are securely handled and stored only for local use within gameplay.
Maze of Me is open to your ideas! Open an issue, fork the repo, or contact the maintainer to contribute.
The license will be finalized before public release. All rights reserved by the author.
🎭 Dive deeper into the labyrinth of your mind. Your journey within Maze of Me awaits.