Advanced
This chapter dives deep into the advanced features of oh-my-opencode: professional AI agent teams, parallel background tasks, Categories and Skills system, lifecycle hooks, and more. Mastering these capabilities will enable you to orchestrate AI workflows like managing a real team, achieving a more efficient development experience.
Chapter Content
AI Agent Teams: 10 Experts Overview
Comprehensive introduction to the functionality, use cases, and calling methods of 10 built-in agents. Learn to select the right agent based on task type, enabling efficient team collaboration, parallel task execution, and deep code analysis.
Prometheus Planning: Interview-Based Requirements Gathering
Clarify requirements and generate work plans through interview mode. Prometheus will continue asking questions until requirements are clear, and automatically consults Oracle, Metis, and Momus to verify plan quality.
Background Parallel Tasks: Work Like a Team
In-depth explanation of the background agent management system usage. Learn concurrency control, task polling, and result retrieval, enabling multiple AI agents to handle different tasks simultaneously, dramatically improving work efficiency.
LSP and AST-Grep: Code Refactoring Tools
Introduction to the usage of LSP tools and AST-Grep tools. Demonstrate how to implement IDE-level code analysis and operations, including symbol navigation, reference lookup, and structured code search.
Categories and Skills: Dynamic Agent Composition
Learn to use the Categories and Skills system to create specialized sub-agents. Implement flexible agent composition, dynamically assigning models, tools, and skills based on task requirements.
Built-in Skills: Browser Automation and Git Expert
Detailed introduction to the use cases and best practices of three built-in Skills (playwright, frontend-ui-ux, git-master). Quickly access professional capabilities like browser automation, frontend UI design, and Git operations.
Lifecycle Hooks: Automated Context and Quality Control
Introduction to the usage of 32 lifecycle hooks. Understand how to automate context injection, error recovery, and quality control, building a complete AI workflow automation system.
Slash Commands: Preset Workflows
Introduction to the usage of 6 built-in slash commands. Including /ralph-loop (quick fix loop), /refactor (code refactoring), /start-work (start project execution), and other common workflows.
Configuration Deep Customization: Agent and Permission Management
Teach users deep customization of agent configuration, permission settings, model overrides, and prompt modifications. Master complete configuration capabilities, creating AI workflows that align with team standards.
Learning Path
┌─────────────────────────────────────────────────────────────────────────────────────────────────────────────┐
│ Step 1 Step 2 Step 3 Step 4 │
│ Understand AI Agents → Master Planning → Learn Dynamic Agent → Deep Customization │
│ (Basic Concepts) & Parallel Tasks Composition & Automation │
│ (Advanced Usage) (Expert Level) │
└─────────────────────────────────────────────────────────────────────────────────────────────────────────────┘Recommended Sequence:
- Start with AI Agent Teams: Understand the responsibilities and use cases of 10 agents—this is the cornerstone for understanding the entire system.
- Then Learn Planning and Parallel Tasks: Master Prometheus planning and the background task system—this is the core of efficient collaboration.
- Next Learn Dynamic Agent Composition: Learn Categories and Skills to achieve flexible agent specialization.
- Finally Learn Deep Customization: Master lifecycle hooks, slash commands, and configuration customization to build complete workflows.
Advanced Routes:
- If your goal is quick start: Focus on "AI Agent Teams" and "Slash Commands"
- If your goal is team collaboration: Deep dive into "Prometheus Planning" and "Background Parallel Tasks"
- If your goal is workflow automation: Learn "Lifecycle Hooks" and "Configuration Deep Customization"
Prerequisites
Before You Begin
This chapter assumes you have completed:
- ✅ Quick Installation and Configuration: Installed oh-my-opencode and configured at least one Provider
- ✅ First Look at Sisyphus: Main Orchestrator: Understood basic agent calling and delegation mechanisms
- ✅ Provider Configuration: Claude, OpenAI, Gemini: Configured at least one AI Provider
Next Steps
After completing this chapter, we recommend continuing with:
- Configuration Diagnostics and Troubleshooting: Quickly locate and resolve issues when they arise.
- Configuration Reference: View the complete configuration file schema and understand all configuration options.
- Claude Code Compatibility: Learn how to migrate existing Claude Code workflows to oh-my-opencode.