Generation Prompt

The skill factory works best when the prompt is structured. The prompt should tell the AI what input it receives, what artifacts it must generate, and which skills it must connect to each issue.

Example:

You are helping decompose a feature into implementation-ready issue files.

Input:

- General functional description
- Available skills
- Project constraints

Generate:

- user stories
- functional breakdown
- `ISSUE-n.md` files
- required skills per issue
- expected output per issue

Rules:

- Keep each issue small.
- Each issue must contain exactly one main user story.
- Each issue must list the functions the implementation should expose or use.
- Each issue must reference `.skills/SKILLS.md`.
- Add topic-specific skills when useful.
- Do not generate implementation code yet.

Give the AI the available skills as explicit context. This prevents it from inventing standards that conflict with the project.

Example:

Available skills:

- `.skills/SKILLS.md`
- `.skills/engineering-skills/react/page.md`
- `.skills/engineering-skills/react/molecule.md`
- `.skills/engineering-skills/quality/testing.md`
- `.skills/engineering-skills/storybook/component-states.md`
- `.skills/domains/product-catalog.md`
- `.skills/features/product-overview.md`

Then provide the functional description and request concrete files. The output should be ready to copy into ISSUE-n.md files after human review.

Example:

Functional description:

Build a product overview page for internal users.
Users can view active products, search by product name or SKU,
filter by stock status, and open product details.

Return the result as separate Markdown sections:

- `ISSUE-1.md`
- `ISSUE-2.md`
- `ISSUE-3.md`
- `ISSUE-4.md`
- `ISSUE-5.md`

A good generated issue connects the story, functions, skills, and output in one place.

Example:

# ISSUE-3.md

## Title

Add product search

## User Story

As an internal user,
I want to search products by name or SKU,
so that I can find a product quickly.

## Functions

- `normalizeProductSearchTerm(value)`
- `filterProductsBySearch(products, term)`
- `renderProductSearchInput(value, onChange)`

## Required Skills

- `.skills/SKILLS.md`
- `.skills/engineering-skills/react/molecule.md`
- `.skills/engineering-skills/quality/testing.md`
- `.skills/domains/product-catalog.md`

## Expected Output

- Search input component
- Search filtering function
- Unit tests for search behavior
- Storybook story for default and filled states

Mini Exercise

Use the prompt structure above with your own functional description. Review whether every generated issue has one story, a small function list, required skills, and expected output.

Mini Exercise​

Mini Exercise