Watsonx Agent Deploy

Simple CLI tool to deploy multiple AI agents to IBM watsonx.ai

Installation

pip install watsonx-agent-deploy

Quick Start

1. Generate agent folders using the IBM CLI template:

   watsonx‑ai template new "base/langgraph‑react‑agent" my-agent
   

2. Create a .env file with:

   WATSONX_APIKEY=your_api_key_here
   WATSONX_URL=https://your-watsonx-url.com
   SPACE_ID=your_space_id
   

3. Deploy agents by running:

   watsonx-deploy
   

Required File Structure

Your project directory (current or specified via --config-dir) should follow this layout:

your-project-root/
├── .env                          # contains WATSONX_APIKEY, WATSONX_URL, SPACE_ID
├── my-agent/
│   └── config.toml              # agent-specific configuration
├── another-agent/
│   └── config.toml
├── sample-agent/
│   └── config.toml

Notes:

• Agent folders must end with 'agent' (e.g. my-agent), else they will be ignored.
• Each agent folder must contain a valid config.toml.
• The .env file must reside in the root of your deployment directory (or be specified via --env-file).

Usage Examples

# Deploy agents from current default directory
watsonx-deploy

# Use a custom .env file
watsonx-deploy --env-file path/to/.env

# Deploy engines from a specific directory
watsonx-deploy --config-dir path/to/agents

# Enable verbose logging for troubleshooting
watsonx-deploy --verbose

Requirements

• Python 3.11+
• Proper .env with required credentials
• Agent folders must follow naming and structure rules above

License

MIT