If you're using OpenAI Codex CLI and want something closer to Claude's multi-agent workflows, this gives you 30+ role prompts (architect, debugger, security-reviewer) and 35+ workflow skills that coordinate them. The standout pieces are $ralph for persistent execution loops, $team for spawning coordinated tmux workers, and $autopilot for full idea-to-code pipelines without leaving Codex. It's basically a runtime layer that adds reusable skills and parallel agent orchestration on top of the base Codex CLI. Start with omx --madmax --high, then pull in $plan or $deep-interview when scope is unclear, or $ultrawork when you need independent agents running in parallel. Cross-model consultation with Claude or Gemini is built in.
npx -y skills add akillness/oh-my-skills --skill omx --agent claude-codeInstalls into .claude/skills of the current project.
Select a file.
juliusbrussee/caveman
mattpocock/skills
shadcn/improve
obra/superpowers
forrestchang/andrej-karpathy-skills
vercel-labs/skills