Agent Execution Commands
Commands for executing individual agents or agent scripts.
rad stepβ
Execute a single agent with optional input.
Usageβ
rad step <id> [prompt...] [options]
Argumentsβ
id- Agent ID from configurationprompt- Additional prompt to append (optional)
Optionsβ
--model <model>- Model to use (overrides agent config)--engine <engine>- Engine to use (overrides agent config)--reasoning <level>- Reasoning effort level (low, medium, high)
Examplesβ
# Execute agent with default prompt
rad step arch-agent
# Execute with additional prompt
rad step code-agent "Implement a REST API endpoint"
# Override model
rad step code-agent --model gpt-4
# Override engine
rad step code-agent --engine claude
# Set reasoning level
rad step code-agent --reasoning high
rad runβ
Execute agent script with parallel (&) and sequential (&&) support.
Usageβ
rad run <script> [options]
Argumentsβ
script- Agent script (e.g., "agent-id 'prompt'" or "agent1 & agent2")
Optionsβ
--model <model>- Model to use-d, --dir <path>- Working directory
Examplesβ
# Run single agent
rad run "code-agent 'Implement feature X'"
# Run agents in parallel
rad run "agent1 'task1' & agent2 'task2'"
# Run agents sequentially
rad run "agent1 'task1' && agent2 'task2'"
# Run with specific model
rad run "agent-id 'prompt'" --model gpt-4
# Run in specific directory
rad run "agent-id 'prompt'" --dir /path/to/project
rad chatβ
Interactive chat mode with session management.
Usageβ
rad chat [agent-id] [options]
Argumentsβ
agent-id- Agent ID to chat with (optional with --list)
Optionsβ
--session <name>- Session name (defaults to timestamp)--resume- Resume an existing session--list- List available sessions
Examplesβ
# Start chat session
rad chat code-agent
# Start named session
rad chat code-agent --session my-session
# Resume session
rad chat code-agent --resume
# List sessions
rad chat --list