Monolith CLI

Command-line tool for the Monolith Computer Vision Platform — manage projects, datasets, annotations, model training, deployment, and multi-model workflows from the terminal.

Install

npm install -g monolith-cli

Supports Linux (x64, arm64), macOS (Intel, Apple Silicon), and Windows (x64).

Quick Start

monolith setup        # configure server + API key (interactive)
monolith org list     # list organizations
monolith project list # list projects

Core Commands

| Task | Command | |------|---------| | Upload data | monolith dataset upload ./images/ | | Auto-label | monolith dataset pre-annotate --backend rexomni --wait | | Review labels | monolith dataset serve | | Split data | monolith dataset split | | Train model | monolith model train --base-model <id> --wait | | Evaluate | monolith model test --test-model <id> --wait | | Deploy | monolith model deploy --backend tensorrt --hardware NV_T4 ... | | Predict | monolith model predict ./test.jpg | | Build pipeline | monolith workflow create --name "pipeline" | | Download model | monolith model download <id> --kind onnx -o ./ |

See monolith --help for all commands and options.

Pre-Annotation Backends

Three tiers of AI backends for auto-labeling:

| Tier | Detection | Classification | Best for | |------|-----------|---------------|----------| | Lite | sam3 | siglip | Fast local inference | | Pro | rexomni | qwen | Balanced accuracy | | Max | doubao | doubao_cls | Highest accuracy (cloud API) |

Plus trained backend to re-use your fine-tuned models.

Docs

• monolith <command> --help — built-in help for every command
• GitLab Repository

License

MIT