customaize-agent-create-agent

To create a custom AI agent in OpenCode, you define the agent's file structure, configure frontmatter fields, and design a system prompt that dictates its autonomous behavior for complex tasks.

Triggering patterns for autonomous agents define the specific conditions for invocation, ranging from explicit user requests to implicit contextual needs, ensuring the agent activates only when required.

System prompt design directly shapes subagent behavior by providing explicit instructions and validation rules, ensuring the AI consistently handles multi-step tasks according to your defined parameters.

Prior prompt engineering experience is helpful but not strictly required, as this process provides AI-assisted generation prompts and best practices to guide you through defining agent purpose and behavior.

The best way to structure an agent file for multi-step tasks is to follow standard validation rules, properly format the frontmatter fields, and write a comprehensive system prompt for independent operation.

Limitations when defining custom agent triggers involve adhering to specific validation rules to prevent conflicts, ensuring your triggering patterns do not overlap excessively with implicit contextual needs.