How many AI models are used in a review?

Depending on the review type, the skill typically coordinates between 2 to 4 models in parallel to ensure a diverse range of perspectives.

Can I use this for existing code implementation?

This skill is designed for the planning phase. For reviewing code already in implementation, the 'sdd-fidelity-review' skill is recommended instead.

Where are the detailed review reports stored?

Full reports are saved locally in the 'specs/.plan-reviews/' directory to keep your conversation history clean and save context tokens.

Does this skill modify my specification files?

No, sdd-plan-review is strictly advisory. it provides structured feedback and recommendations without altering your original spec files.

When should I use the Security review type?

You should use the security review type whenever a specification involves authentication, payment processing, PII handling, or data encryption.

SDD Plan Reviewer

Name: SDD Plan Reviewer
Author: tylerburleigh

bytylerburleigh

Security & Testing

Coordinates multiple AI models to provide structured, multi-dimensional feedback on technical specifications before implementation begins.

About

The sdd-plan-review skill is a specialized quality assurance tool for Spec-Driven Development (SDD) that leverages parallel AI model execution to audit technical plans. By synthesizing diverse perspectives from various AI providers, it identifies architectural flaws, security risks, and feasibility gaps without modifying the original documents. It acts as an automated advisory layer that surfaces consensus findings and prioritized recommendations, helping developers validate complex designs and security-critical implementations before a single line of code is written.

Key Features

Four specialized review types: Quick, Full, Security, and Feasibility
0 GitHub stars
Isolated reporting that saves detailed analysis to Markdown and JSON files to preserve token limits
Structured feedback categorization by priority levels from Critical to Low
Parallel multi-model consultation with automated synthesis of consensus findings
Seamless integration with Foundry MCP for direct specification and provider management

Use Cases

Performing automated peer reviews on draft specs to identify missing requirements before coding
Auditing complex architectural specifications for technical feasibility and hidden dependencies
Validating security-critical designs such as authentication, encryption, and data privacy handling

About

Key Features

Four specialized review types: Quick, Full, Security, and Feasibility
0 GitHub stars
Isolated reporting that saves detailed analysis to Markdown and JSON files to preserve token limits
Structured feedback categorization by priority levels from Critical to Low
Parallel multi-model consultation with automated synthesis of consensus findings
Seamless integration with Foundry MCP for direct specification and provider management

Use Cases

Performing automated peer reviews on draft specs to identify missing requirements before coding
Auditing complex architectural specifications for technical feasibility and hidden dependencies
Validating security-critical designs such as authentication, encryption, and data privacy handling