Where do LLM prompts and test scenarios belong?

Prompts should be stored in a 'prompts/' sub-folder and test scenarios should be placed in a 'scenarios/' sub-folder within the workflow directory.

Why is a standardized folder structure important for Output SDK?

It ensures consistency across the codebase, simplifies developer onboarding, and enables proper automated tooling support for AI workflows.

What are the required files in an Output SDK workflow folder?

Every workflow folder must contain workflow.ts for orchestration, steps.ts for I/O operations, and types.ts for Zod schemas.

Can I share HTTP clients between different workflows?

Yes, shared resources like HTTP clients live in a central 'src/clients/' directory and are imported into specific workflow steps as needed.

How should I name my workflow folders and files?

Workflow folders should use snake_case, while the main TS files (workflow.ts, steps.ts, types.ts) are fixed. Prompt files use camelCase@v{n}.prompt format.

Output SDK Folder Structure

Name: Output SDK Folder Structure
Author: growthxai

bygrowthxai

0•

API開発

Standardizes the organization of Output SDK project files to ensure consistent workflow architecture and tooling support.

This skill provides a comprehensive guide to the standard directory layout for Output SDK workflows, ensuring that projects remain maintainable, scalable, and compatible with automated tooling. It defines the specific roles of core files like workflow definitions, step functions, and Zod schemas, while also establishing naming conventions for folders, prompts, and scenarios. By following these conventions, developers can easily navigate complex AI workflow projects and ensure a proper separation of concerns between deterministic orchestration logic and side-effect-heavy I/O operations.

主な機能

01Standardizes folder structure for Output SDK workflow directories

020 GitHub stars

03Defines mandatory file roles for workflows, steps, and type definitions

04Enforces naming conventions for directories, TypeScript files, and prompts

05Provides organization patterns for LLM prompts and test scenarios

06Ensures architecture compliance for automated tooling and consistency

ユースケース

01Refactoring existing AI workflow projects to comply with standard SDK patterns

02Conducting code reviews to verify project structural integrity and compliance

03Initializing a new Output SDK workflow with the correct file hierarchy

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add growthxai/output-claude-plugins output-dev-folder-structure

For use in Claude.ai and ChatGPT

Download Skill