review-plan

Review LP plan packages for gaps, ambiguities, and risks.

Updated Apr 14, 2026
One-click install
npx skills add https://github.com/daudaudinang/vibecode --skill review-plan-daudaudinang
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: review-plan
Source: https://github.com/daudaudinang/vibecode/tree/main/.agents/skills/review-plan
Command: npx skills add https://github.com/daudaudinang/vibecode --skill review-plan-daudaudinang

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Worker-only skill reviews one canonical LP plan package to validate business intent, UX coherence, architecture scope, and edge cases before implementation.

Core Features & Use Cases

  • Dual-mode execution: Standard mode spawns four independent agents (senior_pm, senior_uiux_designer, senior_developer, system_architecture) to review different perspectives in parallel; Fast mode performs a condensed evaluation with a single agent across all four personas for quicker feedback.
  • Inputs validation: Checks inputs such as .codex/plans/PLAN_<NAME>/plan.md, optional .codex/plans/PLAN_<NAME>/spec.md, and any phase-XX-*.md files, plus manifests like ownership.json and dependency-graph.json when present.
  • Output governance: Produces a formal verdict report and a binding contract to gate orchestrator decisions.
  • Independence and guardrails: Enforces that no plan modifications occur during review and that persona outputs are reconciled for conflicts.

Quick Start

Provide the LP plan package to the skill and start the standard review workflow.

Frequently Asked Questions about review-plan

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

FAQPage Schema
How do I review a product plan package for architecture and UX risks before implementation?

A multi-persona plan review validates business intent, UX coherence, architecture scope, and edge cases before implementation. It checks canonical plan packages and outputs a structured report with a binding contract to gate orchestrator decisions.

What is the difference between standard mode and fast mode for plan validation?

Standard mode spawns four independent agents (senior_pm, senior_uiux_designer, senior_developer, system_architecture) to review perspectives in parallel, while fast mode performs a condensed evaluation with a single agent across all four personas for quicker feedback.

What input files are required to run an LP plan package review?

LP plan package review requires the `plan.md` file located under `.codex/plans/PLAN_<NAME>/`. Optional inputs include `spec.md`, phase-specific markdown files, and manifests like `ownership.json` and `dependency-graph.json` if present.

Can I modify my plan during the risk assessment process?

No, the risk assessment enforces guardrails so no plan modifications occur during review. It operates independently and only outputs a formal verdict report and a binding contract to gate downstream implementation.

Where does the plan review output its structured reports and contracts?

The plan review outputs its formal verdict report and binding contract to the `.codex/pipeline/PLAN_<NAME>/` directory. This structured output reconciles persona conflicts and provides governance for orchestrator decisions.