ralplan

ralplan is a planning entrypoint from the oh-my-claudecode project that runs a consensus planning process before any work starts. It is a shorthand for the plan command with its consensus option, and it runs an iterative loop of a Planner, an Architect and a Critic agent until they agree on a plan, using a structured deliberation method called RALPLAN-DR. The default is a short mode, and a deliberate mode is used for high-risk work, which adds a pre-mortem of three scenarios and expanded test planning. Deliberate mode also turns on automatically when a request signals high risk, such as authentication or security, data migrations, destructive changes, production incidents, compliance or personal data, or changes that break a public API. Flags control the process. The interactive flag adds prompts at the draft review and the final approval, and without it the loop runs unattended, marks the final plan as pending approval, and stops without executing anything. Other flags let the Codex CLI handle the Architect or Critic role when it is available, with a fallback to the default Claude review otherwise. It is useful for teams that want a vague request turned into a reviewed plan before a build or team run begins. Plans produced without the interactive flag still need a person to approve them before execution. The skill is installed with the skills CLI from the yeachan-heo/oh-my-claudecode repository. Its skills.sh page showed about 1.2K installs and passing results from three security audits at the time of the scan.…