What is Veraforge

VeraForge is a compiler designed for the Vera family of languages, including VeraPoetry, VeraDSL, and VeraNL. It targets various executable formats such as fast-agent, MCP, Ollama, dex-lang, and IR.

Use cases

Use cases for VeraForge include developing applications that require multi-modal sensor fusion, such as autonomous vehicles, smart home systems, and industrial automation, where real-time data analysis and decision-making are critical.

How to use

To use VeraForge, you can compile Vera language files using the command line interface (CLI). For example, to compile a file to fast-agent Python, you would run: python veraforge.py compile examples/sensor-fusion-bot.vera --target=fast-agent --output sensor_fusion_bot.py. You can also compile to IR for debugging.

Key features

Key features of VeraForge include support for multiple Vera languages, targeting various executable formats, and a command-line interface for compiling and testing. It also provides a structured IR format for system visualization and debugging.

Where to use

VeraForge can be used in fields such as robotics, sensor data processing, and real-time systems, where integration of multiple sensor inputs and real-time processing is required.

VeraForge

Compiler for the family of Vera languages (VeraPoetry, VeraDSL, VeraNL) targeting an array of executable formats (fast-agent, MCP, Ollama, dex-lang, IR, etc.)

Note: this project is in early design phases.

CLI Compile Command

# Compile to fast-agent Python
python veraforge.py compile examples/sensor-fusion-bot.vera --target=fast-agent --output sensor_fusion_bot.py

# Compile to IR for debugging or system visualization
python veraforge.py compile examples/sensor-fusion-bot.vera --target=ir --output sensor_fusion_bot.vf.json

VeraForgeIR Format (Example)

{
  "name": "SensorFusionBot",
  "essence": [
    "Integrative",
    "Reactive",
    "Real-Time"
  ],
  "purpose": "To fuse multi-modal sensor inputs into a coherent awareness frame for downstream tasks.",
  "channels": {
    "DepthSensor": {
      "type": "Sensor",
      "source": "LIDAR-X5"
    },
    "VisualSensor": {
      "type": "Sensor",
      "source": "VisionCam-A2"
    },
    "AudioSensor": {
      "type": "Sensor",
      "source": "MicArray-V1"
    },
    "AnalysisEngine": {
      "type": "Processor",
      "source": "EdgePredictor"
    }
  },
  "memory": [
    "RecentFrames",
    "ObjectPatterns",
    "NoiseThresholds"
  ],
  "rituals": [
    {
      "name": "FuseInputs",
      "input": [
        "DepthSensor.Grid",
        "VisualSensor.Frame",
        "AudioSensor.Signals"
      ],
      "output": "AnalysisEngine.Frame",
      "process": [
        "Synchronize(Timestamps)",
        "Normalize(DepthSensor, VisualSensor, AudioSensor)",
        "DetectObjects(VisualSensor)",
        "Fuse(DepthSensor, AudioSensor)",
        "OutputTo(AnalysisEngine)"
      ]
    }
  ]
}

CLI Test Command

# Run everything
python veraforge.py test

# Only parsing checks
python veraforge.py test --mode=parse

# Only snapshot verification
python veraforge.py test --mode=snapshot