Multi-agent orchestration framework for AI-driven software development. Coordinates specialized discipline agents (Sisyphus, Hephaestus, Prometheus, Oracle, Librarian, Explore) for task decomposition, parallel execution, and self-improvement loops. Use when orchestrating multiple AI agents, building multi-model workflows, or handling complex development tasks with parallel agents.
Multi-model, multi-agent orchestration framework for software development. Coordinates specialized AI agents working in parallel to solve complex tasks. Published as oh-my-opencode.
Decomposes development tasks into sub-tasks assigned to specialized "Discipline Agents" that execute in parallel across multiple LLMs, with self-verification loops.
| Agent | Role |
|---|---|
| Sisyphus | Repetitive/iterative tasks |
| Hephaestus | Code crafting and implementation |
| Prometheus | Knowledge discovery and learning |
| Oracle | Prediction and analysis |
| Librarian | Documentation and knowledge management |
| Explore | Codebase exploration and navigation |
# Install globally
npm install -g oh-my-opencode
# or
pnpm add -g oh-my-opencode
Supports: Claude, Kimi, GLM-5, GPT-5.4, Gemini, and other LLMs.
# Interactive CLI
oh-my-opencode "Refactor the authentication module to improve security"
# System decomposes task across agents:
# - Explore: Analyze current implementation
# - Oracle: Identify risks
# - Hephaestus: Implement improvements
# - Librarian: Update documentation
# - Ralph Loop: Verify changes
{
"models": {
"primary": "claude-opus",
"fallback": "gpt-4"
},
"agents": {
"parallelism": 4,
"timeout": 300
},
"mcp": {
"enabled": true,
"services": ["web-search", "documentation"]
}
}