Skip to main content

CLI Reference

Usage​

vibex <file.vibe> [flags]

Flags​

FlagDefaultDescription
-o <output>a.outOutput binary name
--arch <arch>arm64Target architecture: arm64 or x86_64
--model <model>gpt-4oOverride the LLM model
--workoffKeep temp directory and print its path

Environment Variables​

Configure via .env in your project directory, ~/.vibex.env globally, or as shell env vars.

VariableDefaultDescription
VIBEX_BASE_URLhttps://api.openai.com/v1OpenAI-compatible API endpoint
VIBEX_API_KEY—API key (OPENAI_API_KEY also works)
VIBEX_MODELgpt-4oModel to use as compiler
VIBEX_ARCHarm64Default target architecture

Examples​

# Basic usage
vibex hello.vibe -o hello

# Use a specific model
vibex hello.vibe -o hello --model gpt-4o

# Compile for x86_64 (requires Rosetta on Apple Silicon)
vibex hello.vibe -o hello --arch x86_64

# Keep temp files for debugging
vibex hello.vibe -o hello --work

# Use a local model
VIBEX_BASE_URL=http://localhost:11434/v1 VIBEX_MODEL=llama3.2 VIBEX_API_KEY=ollama \
vibex hello.vibe -o hello

Inspecting Generated Assembly​

Use --work to keep the temp directory:

vibex hello.vibe -o hello --work
# [vibex] WORK=/var/folders/.../vibex_work_xxx/
# assembly is at /var/folders/.../vibex_work_xxx/program.s