utility-pm-critic

Run adversarial review on PM artifacts with P0-P3 graded findings and fix suggestions.

640|81|Updated Jan 9, 2026
One-click install
npx skills add https://github.com/product-on-purpose/pm-skills --skill utility-pm-critic
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: utility-pm-critic
Source: https://github.com/product-on-purpose/pm-skills/tree/main/skills/utility-pm-critic
Command: npx skills add https://github.com/product-on-purpose/pm-skills --skill utility-pm-critic

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

PM artifacts like PRDs, OKR sets, and personas often ship with hidden gaps, inconsistencies, and weak assumptions that only surface after engineering starts. This Skill stress-tests any PM artifact before it ships, returning severity-graded findings with concrete fixes.

Core Features & Use Cases

  • Adversarial Review: Dispatches the pm-critic sub-agent on Claude Code, or executes its system prompt inline on non-Claude clients like Codex CLI, Cursor, Windsurf, Copilot, and Gemini CLI.
  • Graded Findings: Returns findings graded P0/P1/P2/P3, each with location, issue, impact, and a concrete fix suggestion, plus a machine-readable Status YAML block.
  • Use Case: After drafting a PRD with the deliver-prd skill, run this Skill on the artifact to catch ship-blockers like missing acceptance criteria or inconsistent success thresholds before engineering kickoff.

Quick Start

Ask the AI to run an adversarial review on your PRD file and report every P0 and P1 issue with fixes before you send it to engineering.

Frequently Asked Questions about utility-pm-critic

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

FAQPage Schema
How do I run an adversarial review on a PRD before shipping it?

Invoke the skill with the path to your PRD artifact. It reviews the document against the deliver-prd standards and returns findings graded P0 through P3, each with a concrete fix suggestion and a recommended next action.

What types of PM artifacts can be reviewed for gaps and weak assumptions?

The skill reviews PRDs, OKR sets, personas, lean canvases, meeting recaps, interview syntheses, problem statements, hypotheses, edge case catalogs, and retrospectives. It reads the relevant standards doc for each artifact type before producing findings.

Does this work on AI clients other than Claude Code?

Yes. On Claude Code it dispatches to the native pm-critic sub-agent; on Codex CLI, Cursor, Windsurf, Copilot, and Gemini CLI it reads the pm-critic system prompt and executes the review inline. Production support covers Claude Code and Codex CLI, with other clients experimental.

When should I not use adversarial artifact review?

Do not use it to author new artifacts, run repo-level structural audits, enforce style rules, or review code. Authoring belongs to phase skills like deliver-prd, repo audits belong to the skill auditor, and style enforcement is the release conductor's G0 gate.

What does the review output look like?

Output has three sections: a full findings report with P0/P1/P2/P3 items including location, issue, impact, and fix; a prose Status Summary with a recommended next action; and a machine-readable Status YAML block for tooling or parent agents.