A streamlined, production-ready AI platform that connects your iPhone to local AI models via voice commands.

# Install TinyIntent (one-time setup)
./install.sh

# Start TinyIntent server
tinyintent

# Your server is now running at http://YOUR_IP:8787
# Ready for iPhone voice shortcuts!

1. Open Shortcuts app on your iPhone

2. Create new shortcut with these actions:

   • Dictate Text (Stop Listening: After Pause)
   • Get Contents of URL:
     • URL: http://YOUR_IP:8787/shortcut/route
     • Method: POST
     • Headers: X-Shortcut-Token: tinyintent-shortcut-token-123
     • JSON Body: {"text": "[Dictated Text]", "return_format": "text"}
   • Get Text from Contents of URL (extract speak field)
   • Speak Text

3. Test: Say "Show me system logs" and hear the response!

tinyintent              # Start server (default: port 8787)
tinyintent --port 9000  # Start on custom port
tinyintent status       # Show system status
tinyintent --help       # Show all options

tinyintent/
├── tinyintent/           # Main CLI package
│   ├── cli.py           # Streamlined CLI entry point
│   ├── config.py        # Configuration management
│   └── simple_server.py # Fallback server
├── bridge/              # FastAPI service core
│   ├── routes/          # Modular route organization
│   │   ├── shortcut.py  # iPhone Shortcut API (M11.0)
│   │   ├── health.py    # Health check endpoints
│   │   ├── helpers.py   # Helper management
│   │   ├── agents.py    # Agent lifecycle
│   │   └── system.py    # System operations
│   ├── tinyrpc.py       # Main FastAPI app
│   ├── security.py      # Authentication & authorization
│   └── provenance.py    # Agent signing & tamper detection
├── router/              # SmallIntent.mlmodel routing
│   ├── SmallIntent.mlmodel  # CoreML intent classifier
│   ├── train_router.swift  # Model training
│   └── data/            # Training datasets
├── helpers/             # Sandboxed task execution
│   ├── bot_guard/       # Crypto trading helper
│   ├── log_tailer/      # System log analysis
│   ├── registry.py      # Helper discovery
│   └── executor.py      # Sandboxed execution
├── data/episodes/       # Episode logging & storage
├── tests/               # Comprehensive test suites
├── scripts/             # Utility & automation scripts
├── docs/                # Documentation
├── install.sh           # One-command installation
└── pyproject.toml       # Python packaging

• 🧠 SmallIntent.mlmodel - Local CoreML intent routing
• 📱 iPhone Shortcut Integration - Voice commands via Siri
• 🔒 Security Framework - Authentication, sandboxing, audit logging
• 🤖 Helper System - Extensible action execution
• 📊 System Monitoring - Health checks and metrics
• 🌐 Tailscale Ready - Works anywhere with secure networking

• Endpoint: /shortcut/route
• Auth: X-Shortcut-Token header
• Features: Voice-optimized text formatting, TTS optimization
• Integration: Seamless Siri voice command processing

• Model: SmallIntent.mlmodel (CoreML, runs on Neural Engine)
• Classification: gen (generative) vs act (action execution)
• Training: Swift + CreateML → CoreML artifacts
• Performance: Sub-10ms inference on Apple Silicon

• Runtime: Node.js sandboxed execution with capability isolation
• Security: CPU/memory limits, filesystem restrictions, network controls
• Registry: YAML-based discovery with lifecycle management
• Available Helpers:
  • bot_guard - Crypto trading position management
  • log_tailer - System log analysis and monitoring

Try saying these to your iPhone:

• "Show me recent error logs"
• "Check system health"
• "What's my server status?"
• "Get my trading positions"

# Run comprehensive test suite
make test

# Individual test suites
./tests/health.sh         # System health checks
./tests/auth.sh           # Authentication testing
./tests/routes.smoke.sh   # API endpoint validation
./tests/router_smoke.sh   # Router model testing
./tests/helpers_smoke.sh  # Helper framework testing

# Advanced commands
tinyintent --reload       # Development mode with auto-reload
make doctor              # System diagnostic report
make router-train        # Train new intent classification model

Environment variables (auto-configured with defaults):

TINYINTENT_PORT=8787                              # Server port
SHORTCUT_TOKEN=tinyintent-shortcut-token-123      # iPhone auth token
TINYINTENT_EXECUTION_ENABLED=1                    # Enable helper execution
TINYINTENT_SECRET=your-secret-here                # API authentication

• GET /health - Health check
• GET /shortcut/ping - iPhone Shortcut health check
• POST /shortcut/route - Voice command routing
• GET /docs - Interactive API documentation
• GET /helpers - Available helpers list
• POST /helpers/{id}/preview - Helper preview mode

• 🔒 Local-First: All inference runs locally (no cloud calls)
• 🛡️ Sandboxed Execution: Helpers run in isolated environments with CPU/memory limits
• 🔑 Multi-Layer Auth: Token-based authentication for iPhone + API secret for advanced access
• 📋 Audit Logging: Full request/response/error logging with tamper detection
• 🚦 Lifecycle Gates: Agent staging, approval workflows, and emergency kill switches
• 🔐 Provenance Tracking: Cryptographic signing and tamper-evidence for all agents

iPhone (Siri) → Shortcuts → TinyIntent Bridge → SmallIntent.mlmodel → Helpers → Response

"Command not found": Run ./install.sh first
"Connection refused": Check firewall and network settings
"Invalid token": Verify X-Shortcut-Token header matches configuration
"Module not found": Ensure you're in the TinyIntent project directory

• README_STREAMLINED.md - Quick start guide
• docs/iOS_SHORTCUTS.md - iPhone Shortcut setup
• PRD.md - Complete product requirements
• CLAUDE.md - AI assistant context

# System diagnostics
make doctor

# Security operations  
./scripts/rotate_secret.sh      # Rotate API secret
./scripts/print_urls_and_secret.sh  # Show current config

# Data management
./scripts/export_episodes.sh    # Export training data

TinyIntent v2.0.0 - Built with ❤️ for voice-first AI interaction

Note: This is a local-first platform. No data leaves your Mac.