How does this skill prevent code regressions?

It enforces a 'research before writing' phase where it establishes a passing test baseline and maps the 'blast radius' to identify all affected projects and shared dependencies.

Can I use this for non-monorepo projects?

Absolutely. While it includes specific safety protocols for monorepos, the structured planning and verification workflow benefits any software development project.

What is the Boundary System?

It is a safety framework categorized into 'Always Do', 'Ask First', and 'Never Do' tiers, ensuring the AI agent knows exactly when it needs human intervention for sensitive tasks.

What are machine-verifiable criteria?

These are acceptance criteria written in a GIVEN/WHEN/THEN format that include a specific command (like a test or API call) that Claude can run to verify the result autonomously.

AI PRD Architect

Name: AI PRD Architect
Author: TYLANDER

byTYLANDER

0•

Productivity & Workflow

Generates machine-verifiable Product Requirements Documents and technical plans optimized for AI-assisted development workflows.

The PRD skill transforms high-level feature requests into precise, actionable specifications designed specifically for Claude Code agents. By enforcing codebase research, blast radius analysis, and a test-first methodology, it ensures that new features are implemented with zero regressions. It provides a structured framework—including success criteria with specific verification commands and clear operational boundaries—to bridge the gap between human intent and reliable AI execution, making it particularly effective for complex monorepo environments.

Key Features

01Machine-verifiable acceptance criteria with executable verification commands.

020 GitHub stars

03Progressive decomposition from high-level specs to dependency-ordered task lists.

04Structured boundary system defining autonomous vs. human-approval AI actions.

05Automated blast radius and dependency mapping to prevent cross-project regressions.

06Integrated monorepo safety protocols with mandatory testing baseline enforcement.

Use Cases

01Standardizing technical specifications for AI-driven code generation.

02Planning new feature implementations in complex monorepo environments.

03Analyzing the impact of library changes on downstream consumers.

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add tylander/claude-dev-workflow prd

For use in Claude.ai and ChatGPT

Download Skill