OHMS CLI

The official command-line interface for OHMS AI Foundry - deploy and manage AI agents on Internet Computer.

Installation

npm install -g ohms-cli

Quick Start

1. Configure the CLI (first time only):

   ohms config

2. Deploy OHMS platform (if not already deployed):

   ohms deploy

3. Upload and deploy an AI model:

   ohms ingest model.onnx --name "my-model" --type "text"

4. List deployed agents:

   ohms list

5. Run inference:

   ohms predict <agent-id> "Hello, world!"

Commands

ohms config

Configure OHMS CLI settings including network and canister IDs.

Options:

• --network <network> - Target network (local, mainnet)
• --registry <id> - Registry canister ID
• --orchestrator <id> - Orchestrator canister ID

ohms deploy

Deploy OHMS platform canisters to Internet Computer.

Options:

• -n, --network <network> - Target network (local, ic) (default: local)
• -r, --reinstall - Reinstall canisters (destroys data)
• -u, --upgrade - Upgrade existing canisters

ohms ingest <model-path>

Upload and deploy an ONNX model as an agent.

Arguments:

• model-path - Path to ONNX model file

Options:

• -n, --name <name> - Agent name
• -t, --type <type> - Model type (text, image, code)
• -d, --description <desc> - Agent description
• --public - Make agent publicly accessible
• --skip-conversion - Skip ONNX to WASM conversion (for testing)

ohms list

List deployed agents.

Options:

• -t, --type <type> - Filter by agent type (text, image, code)
• -f, --format <format> - Output format (table, json) (default: table)
• --filter <text> - Filter agents by name or description

ohms predict <agent> [input]

Run inference on a deployed agent.

Arguments:

• agent - Agent ID or name
• input - Input data for prediction

Options:

• -f, --format <format> - Input format (text, json) (default: text)
• -o, --output <format> - Output format (text, json) (default: text)

Examples

Deploy a Text Model

# Upload and deploy a language model
ohms ingest tinyllama.onnx --name "tinyllama" --type "text" --description "Tiny Llama language model"

# Run inference
ohms predict tinyllama "What is the capital of France?"

Deploy an Image Model

# Upload and deploy an image classification model
ohms ingest mobilenet.onnx --name "mobilenet" --type "image" --description "MobileNet image classifier"

# Run inference (input as base64 or file path)
ohms predict mobilenet "data:image/jpeg;base64,/9j/4AAQSkZJRgABAQAAAQ..."

Deploy to Mainnet

# Deploy platform to mainnet
ohms deploy --network ic

# Configure for mainnet
ohms config --network mainnet

# Upload model to mainnet
ohms ingest model.onnx --name "production-model"

Requirements

• Node.js 18.0.0 or higher
• dfx (DFINITY Canister SDK) for deployment
• Internet Computer identity (for mainnet)

Development

# Clone the repository
git clone https://github.com/OHMS-DeAI/ohms-core.git
cd ohms-core/packages/cli

# Install dependencies
npm install

# Build the CLI
npm run build

# Run in development mode
npm run dev

# Test the CLI
node dist/index.js --help

Support

• Documentation: https://ohms.ai/docs
• Issues: https://github.com/OHMS-DeAI/ohms-core/issues
• Discord: https://discord.gg/ohms

License

MIT License - see LICENSE for details.