spec-review

Review technical specification markdown files for missing requirements, contradictions, and edge-case ambiguities.

Updated May 17, 2026
One-click install
npx skills add https://github.com/krzyssikora/claude-personal-plugin --skill spec-review-krzyssikora
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: spec-review
Source: https://github.com/krzyssikora/claude-personal-plugin/tree/main/plugins/spec-review/skills/spec-review
Command: npx skills add https://github.com/krzyssikora/claude-personal-plugin --skill spec-review-krzyssikora

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

It identifies gaps, ambiguities, contradictions, and missing requirements in a specification so implementation can be planned and executed with fewer surprises.

Core Features & Use Cases

  • Iterative fresh-context review loop: Runs up to a guarded number of rounds using a reviewer subagent and your accepted/rejected feedback to converge on a clean verdict.
  • Spec-genre tuned analysis: Ensures the reviewer is focused on specs under docs/superpowers/specs/ and surfaces issues appropriate to technical requirements documentation.
  • Git-safe workflow and repeatability: Verifies the spec is in a clean, branch-checked-out git state, applies edits at the right anchored locations, and commits each successful round with traceable messaging.

Quick Start

Use the spec-review skill after brainstorming to review the newest spec under docs/superpowers/specs/ and return a verdict plus prioritized catch list.

Frequently Asked Questions about spec-review

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

FAQPage Schema
How do I review a technical specification for missing requirements and ambiguities?

Technical spec review identifies gaps, contradictions, and missing requirements in specification markdown files using an iterative fresh-context loop, returning a prioritized catch list and clean verdict.

Can I run an automated spec review on markdown files under a specific docs directory?

Yes, automated spec review targets markdown files under docs/superpowers/specs/, resolving the target path and dispatching a structured reviewer prompt to analyze technical requirements documentation.

What is the best way to iteratively review specifications after brainstorming outputs?

Iterative spec review runs guarded rounds using a reviewer subagent with accepted/rejected feedback to converge on a clean verdict, applying safe edits until reaching a clean state or round caps.

Does the spec review process require a clean git state before running?

Yes, spec review enforces git preconditions by verifying the specification is in a clean, branch-checked-out git state to ensure a reliable baseline before applying and committing iterative edits.

How does the spec review loop handle stuck states or unresolved ambiguities?

The agentic loop applies and commits safe edits iteratively until reaching a clean verdict, detecting a stuck state, or hitting round caps, preventing infinite loops while maximizing specification clarity.

What limitations exist when reviewing edge-case ambiguities in technical specifications?

Spec review is limited to markdown files under docs/superpowers/specs/ and relies on round caps to prevent infinite loops, stopping at a clean verdict, stuck state, or maximum iteration limit.