AI-powered waifu companion for developers
Choose a waifu. Chat in character. Let her become your coding agent.
An AI-powered companion app where users pick a personalized "waifu" through a dating-style interface. Each waifu has a unique personality, tone, and communication style. Once matched, the waifu becomes an intelligent assistant that lives on your device (desktop or mobile), capable of executing real tasks — from writing code and managing files to automating workflows.
Every interaction is delivered in-character, blending emotional engagement with real productivity.
Current status: see STATE.md for an accurate snapshot of what's built. Earlier status documents have been moved to
docs/archive/for historical reference only.
- 20 AI providers registered, 18 live — Anthropic, OpenAI, OpenAI Codex, Gemini, Mistral, Cohere, Groq, DeepSeek, Perplexity, Together, xAI (2 endpoints), Hugging Face, GitHub Models, MiniMax (global + CN), Ollama, LM Studio. Azure OpenAI and Fireworks are registered but currently throw "not yet fully implemented" (see STATE.md)
- Waifu personalities — distinct characters with unique tones and backstories
- Agent modes — ask / auto / full; plan-and-verify workflow with iteration-budget feedback
- Agent tools — terminal (with destructive-pattern gating), read/write/edit_file, clipboard, git_status / git_diff, web_search, todo_write, Spotify
- Affection system — relationship meter that evolves with conversation
- AI memory — persistent memory across all chats
- Theming — 17 presets including live rainbow hue-cycle and sakura-petals overlay
- Chat niceties — image attachments (paste / drop / paperclip), token + cost counter, per-message regenerate/delete, Markdown export, keyboard shortcuts, tray icon + global shortcut
- Desktop & mobile — Electron desktop app with QR pairing to React Native mobile
- Runtime ops — health probes, Prometheus metrics, Grafana dashboards, container builds, and Kubernetes manifests
- Backups — export and restore chat + memory snapshots from the runtime API
- Plugin system — manifest-based tool plugins for extending agent capabilities
syntax-senpai/
├── apps/
│ ├── desktop/ # Electron + Vite desktop application
│ ├── mobile/ # React Native mobile app (Expo)
│ └── runtime/ # Node.js runtime service for health, metrics, backups, plugins
├── packages/
│ ├── ai-core/ # AI provider abstraction (Anthropic, OpenAI, DeepSeek, etc.)
│ ├── waifu-core/ # Waifu personalities, system prompts, expressions
│ ├── agent-tools/ # Tool execution framework for agents
│ ├── storage/ # Persistent storage layer
│ ├── ui/ # Shared UI components and styles
│ ├── ui-transitions/ # Reusable transition components
│ ├── ui-loading-screens/ # Loading screen components
│ └── ws-protocol/ # WebSocket protocol definitions
├── ops/ # Prometheus & Grafana configuration
├── plugins/ # External tool plugins directory
└── docker-compose.yml # Local dev environment setup
- Node.js >= 20
- pnpm >= 8
# Clone the repository
git clone https://github.com/404-Waifu-Not-Found/SyntaxSenpai.git
cd SyntaxSenpai
# Install dependencies
pnpm install
# Run the desktop app in dev mode
pnpm dev:desktop
# Or quick-start (skips full turbo pipeline)
pnpm desktop:quickAPI keys are configured in-app through the Settings panel — no .env file required for basic usage.
| Command | Description |
|---|---|
pnpm dev:desktop |
Start the desktop app in dev mode |
pnpm dev:mobile |
Start the mobile app in dev mode |
pnpm dev:runtime |
Start the runtime service locally |
pnpm build |
Build all workspaces |
pnpm test |
Run tests |
pnpm lint |
Run linters |
pnpm typecheck |
TypeScript type checks |
pnpm docker:build |
Build the runtime container image |
pnpm docker:up |
Start runtime + Prometheus + Grafana |
pnpm clean |
Remove all build artifacts and node_modules |
The runtime service exposes:
/healthzfor liveness checks/readyzfor readiness checks/metricsfor Prometheus scraping/api/v1/backups/*for chat and memory backup management/api/v1/pluginsfor plugin discovery
See ops/README.md for Docker, monitoring, backup, plugin, and Kubernetes usage.
See CONTRIBUTING.md for development setup, guidelines, and pull request process.
See SECURITY.md for reporting vulnerabilities.
MIT — see the LICENSE file for details.