The AI Workbench is a Faber CLI application for local model inventory, metadata inspection, embedding, indexing, and inference workflows. It demonstrates Faber building a substantial multi-command CLI application with real I/O, JSON output, and Python harness validation.
Package
examples/ai-workbench/packages/faber-ai/ with CLI subcommands:
model inspect— query local model aliases, routes, and statusembed— generate embeddings from text input
Commands
faber check examples/ai-workbench/packages/faber-ai
faber test examples/ai-workbench/packages/faber-ai
faber run examples/ai-workbench/packages/faber-ai -- model inspect basic/minilm --format jsonValidation
The AI Workbench includes 20+ Python harness scripts that compare Faber output against fixture maps for model inventory, inference, GPU evidence, session lifecycle, and package reuse — demonstrating cross-language validation of compiled Faber binaries.