PART 7 · MODULE 7Batch, Teams, Ralph & more
OMC Workflows
Advanced orchestration workflows for parallel and coordinated AI development
Loading module...
Advanced orchestration workflows for parallel and coordinated AI development
OMC (oh-my-claudecode)is an orchestration layer that turns Claude Code into a full development team. Think of it as a "team manager" that coordinates multiple AI agents, each with a specialty, to tackle complex tasks together.
Where Claude Code is a single skilled developer, OMC gives you an entire engineering organization — with project managers, architects, coders, reviewers, and testers.
Need to make changes across many files at once? /batch breaks a large task into independent units, then spawns separate worker agents that all run in parallel.
Each worker gets its own isolated copy of the codebase (a git worktree), so they never step on each other's toes. When they finish, each one creates a pull request.
/batch — Parallel Work Orchestration
When tasks depend on each other, you need coordination, not just parallelism. /team creates a shared task list and assigns multiple agents to work through it together.
Agents pick up tasks, coordinate through the shared list, and can hand off work to each other. One agent might build the backend while another builds the frontend that consumes it.
/team — Coordinated Multi-Agent Teams
RALPH stands for Recursive Adaptive Learning and Planning with Human-feedback. It's a self-referential loop: plan, implement, verify. If the task isn't complete, iterate. Keep going until a verifier confirms everything passes.
Human feedback checkpoints let you steer the process. Ralph never gives up — it keeps finding and fixing issues until the job is truly done.
/ralph — Recursive Loop Until Complete
Keeps iterating until the verifier confirms the task is complete
Give it a goal, and /autopilot handles everything: understanding the goal, planning the approach, implementing code, running tests, and verifying the result. Minimal human intervention needed.
Best for well-defined tasks you trust the AI to handle end-to-end without hand-holding.
A parallel execution engine designed for many similar, independent tasks. Think of it as the "assembly line" mode — optimized for churning through repetitive work fast.
CCG sends the same question to Claude, Codex, and Gemini simultaneously, then synthesizes the best answer from all three. Three expert opinions for the price of one question.
/plan enters a strategic planning mode: explore the codebase, understand the constraints, and design an approach before writing a single line of code.
/ralplan goes further — it auto-detects vague requests and gates them into planning mode before execution. No more jumping into code before understanding the problem.
OMC routes work to the right specialist. Each agent type has deep expertise in its area.
OMC turns a single AI assistant into a full development team. Different workflows for different types of work: /batch for parallel independence, /team for coordination, /ralph for persistence, /autopilot for autonomy. Pick the right tool for the job, and one person can do the work of an entire team.