How are comments added to the document without causing conflicts?

Reviewers write to temporary JSON files. A single merge process then reads all files and inserts blockquote comments into the document at once, preventing parallel editing issues.

How does the parallel review process work?

The skill spawns four independent sub-agents—Maintainability, Security, Efficiency, and CompletenessReviewer—who each generate specialized feedback before the system merges all findings into the document in a single pass.

Can it track resolved comments across iterations?

Yes, the skill can detect existing OPEN or RESOLVED statuses in the document, allowing agents to reopen issues or update existing feedback during re-reviews.

Does it support both specs and implementation plans?

Yes, it automatically detects the mode based on the filename, handling both standard high-level .md specs and detailed -impl.md implementation plans.

What happens if one of the four reviewer agents fails?

The skill logs a warning for the failed agent but proceeds with the remaining agents to ensure the document is updated with as much feedback as possible.

Technical Spec Multi-Agent Reviewer

Name: Technical Spec Multi-Agent Reviewer
Author: shouenlee

byshouenlee

0•

安全与测试

Executes a comprehensive, four-perspective automated review of technical specifications and implementation plans to ensure architectural integrity.

The Spec Review skill automates the rigorous process of vetting technical documentation by orchestrating four specialized AI sub-agents: Maintainability, Security, Efficiency, and Completeness. Designed for complex software engineering workflows, it analyzes technical specs or implementation plans and inserts structured feedback directly into the document. By using a parallel review and single-pass merge architecture, it avoids file conflicts and provides a detailed report of open issues categorized by severity, allowing developers to catch design flaws before coding begins.

主要功能

01Parallel four-agent review (Maintainability, Security, Efficiency, and Completeness)

02State-aware context gathering from ticket metadata and project state files

03Conflict-free comment merging into Markdown specifications

04Automatic mode detection for high-level specs and detailed implementation plans

050 GitHub stars

06Severity-based reporting for Critical, High, Medium, and Low issues

使用场景

01Automated architectural auditing of technical specifications before development starts

02Standardizing documentation quality across engineering teams through AI-driven peer review

03Security and efficiency sanity checks for complex system implementation plans

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add shouenlee/ghcp-dev-plugin spec_review

For use in Claude.ai and ChatGPT

主要功能

01Parallel four-agent review (Maintainability, Security, Efficiency, and Completeness)

02State-aware context gathering from ticket metadata and project state files

03Conflict-free comment merging into Markdown specifications

04Automatic mode detection for high-level specs and detailed implementation plans

050 GitHub stars

06Severity-based reporting for Critical, High, Medium, and Low issues

使用场景

01Automated architectural auditing of technical specifications before development starts

02Standardizing documentation quality across engineering teams through AI-driven peer review

03Security and efficiency sanity checks for complex system implementation plans