userstory to code

The problem

LLMs often generate "hallucinated" code that ignores your project's specific architecture, naming conventions, and existing utilities. This results in generic snippets that require heavy manual refactoring to actually fit your codebase.

What it does

• Analyzes your existing repository to extract real naming styles, export patterns, and directory structures.
• Identifies existing abstractions and shared utilities to prevent duplicate helper functions.
• Maps every individual acceptance criterion from a ticket to a specific code path before implementation.
• Enforces a "plan-first" checkpoint, requiring approval of the architectural approach before writing a single line of code.
• Generates production-ready implementation and matching tests that mirror your team's exact coding style.

Frameworks & tools

Compatible with any stack. It uses grep, glob, and filesystem exploration to adapt to your specific language and framework conventions.

Why this beats prompting it yourself

Generic prompts result in generic code. This skill forces a two-stage verification process that ensures the output isn't just functional, but architecturaly consistent. It eliminates the "copy-paste-refactor" cycle by reading your code before it writes its own.

Use cases

• Translating Jira or Linear tickets into feature stubs and logic.
• Implementing new API endpoints that match existing controller patterns.
• Building UI components that reuse your project's specific design tokens and hooks.
• Converting Gherkin acceptance criteria into functional code and tests.

Known limitations

Requires an existing codebase or defined conventions to match style. It is not designed for bug fixes or pure refactoring tasks without new acceptance criteria.