ralplan

Gate vague multi-agent prompts through Planner, Architect, and Critic consensus reviews.

1|Updated Apr 1, 2026
One-click install
npx skills add https://github.com/OliverOuyang/shuhe-work-skills --skill ralplan-oliverouyang
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: ralplan
Source: https://github.com/OliverOuyang/shuhe-work-skills/tree/main/skills/ralplan
Command: npx skills add https://github.com/OliverOuyang/shuhe-work-skills --skill ralplan-oliverouyang

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Ralplan prevents heavy multi-agent execution from starting on underspecified or vague requests by routing them through a structured consensus planning workflow that produces explicit scope, testable acceptance criteria, and an agreed implementation approach.

Core Features & Use Cases

  • Consensus Gate: Automatically intercepts ralph/autopilot/team style prompts that lack concrete anchors and redirects them to a consensus planning flow.
  • Structured Deliberation: Runs a Planner → Architect → Critic sequential loop with a RALPLAN-DR summary, supports deliberate mode with pre-mortem and expanded test plans, and enforces iterative re-review up to five iterations.
  • Interactive Approval & Bypass: Supports --interactive for user review/approval, --deliberate for high-risk work, flag-based Architect/Critic backends, and explicit bypass signals (e.g., force: or !) when the request is concrete.
  • Use Case: Turn vague requests like "ralph add authentication" into a bounded PRD, acceptance tests, and an approved execution path (team or ralph) before any agents implement changes.

Quick Start

Run /oh-my-claudecode:ralplan with your task description to scope the work and produce a consensus plan before execution.

Frequently Asked Questions about ralplan

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I prevent multi-agent execution from starting on vague requests?

Gate vague multi-agent execution requests through a consensus planning workflow that enforces sequential Planner, Architect, and Critic reviews before implementation. This structured deliberation produces explicit scope, testable acceptance criteria, and an agreed execution path.

What is consensus planning in multi-agent orchestration?

Consensus planning in multi-agent orchestration runs a sequential Planner, Architect, and Critic review loop to resolve vague requests. It generates a summary with explicit scope, architectural review, risk analysis, and testable acceptance criteria before any agents implement changes.

How do I add a planning gate to autopilot or team prompts?

Apply a consensus planning gate to autopilot or team prompts by intercepting requests that lack concrete anchors and redirecting them to a structured deliberation workflow. The gate enforces iterative re-reviews up to five iterations before approving execution.

Can I bypass the consensus planning workflow for concrete prompts?

Yes, you can bypass the consensus planning workflow for concrete prompts by using explicit bypass signals like force: or !. These flags indicate the request already has sufficient detail and skips the Planner, Architect, and Critic review loop.

Does multi-agent consensus planning support interactive approvals and pre-mortem analysis?

Multi-agent consensus planning supports interactive approvals with the --interactive flag and pre-mortem analysis with the --deliberate flag. The deliberate mode expands test plans and risk analysis for high-risk work before implementation begins.

What are the limitations of gating multi-agent execution with consensus planning?

Gating multi-agent execution with consensus planning limits iterative re-reviews to a maximum of five iterations. If consensus is not reached within this loop limit, the structured deliberation cannot automatically approve the execution path for the agents.