project-creator — Claude Code skill: what it does, how to install
project-creator scaffolds structured technical documentation in a .project/ directory and guides you through specs, architecture, and user stories.
MIT 113 installs ★ 23 updated 2026-10-24
npx skills add itechmeat/llm-code --skill project-creator -a claude-codeWhat project-creator does
The skill builds documentation in a set sequence: about.md, specs.md, architecture.md, project-context.md, and then individual user stories in stories/. It does not invent details on its own. Instead, it pulls guidelines from references/, asks focused questions one topic at a time, and offers two or three phrasing options.
Every file uses a Markdown template from the assets/ folder. When documentation already exists, the assistant inspects completion, flags gaps, and asks whether to edit or move forward. Under its workflow rules, the agent will never advance without your explicit confirmation and blocks version downgrades in specs.
What it helps with
- scaffold a
.project/documentation tree - draft technical specifications through guided questions
- map out system architecture and component interactions
- structure feature backlogs into numbered user story files
Example prompts for the agent
- Initialize the
.project/documentation structure for this repository - Guide me through drafting the architecture and technical specs
- Break down our MVP features into user stories in
stories/
How to install project-creator
1Skills CLI (Vercel)
npx skills add itechmeat/llm-code --skill project-creator -a claude-code2Manually: put the files in place
~/.claude/skills/project-creator/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/project-creator/
1Skills CLI (Vercel)
npx skills add itechmeat/llm-code --skill project-creator -a codex2Manually: put the files in place
.agents/skills/project-creator/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add itechmeat/llm-code --skill project-creator -a cursor2Manually: put the files in place
.cursor/skills/project-creator/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install itechmeat/llm-code skills/project-creator2Manually: put the files in place
.github/skills/project-creator/SKILL.md1Skills CLI (Vercel)
npx skills add itechmeat/llm-code --skill project-creator -a gemini-cli2Platform CLI
gemini skills install https://github.com/itechmeat/llm-code --path skills/project-creatorManually: put the files in place
.devin/skills/project-creator/SKILL.mdThe legacy .windsurf/skills/ path still works
Skills CLI (Vercel)
npx skills add itechmeat/llm-code --skill project-creator -a opencodeClaude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about project-creator
How do I install project-creator in Claude Code?
Run npx skills add itechmeat/llm-code --skill project-creator -a claude-code in your terminal. This writes the skill definition to ~/.claude/skills/project-creator/SKILL.md.
What does the project-creator skill do?
It guides you through an interview to populate about.md, specs.md, architecture.md, and user stories. The assistant uses templates and requests confirmation before saving drafts.
In what order are the documentation files created?
It follows a strict pipeline: about.md leads into specs.md, followed by architecture.md, project-context.md, and final user stories in stories/. The agent completes each file before moving to the next.
Source text
SKILL.md · MIT
---
name: project-creator
description: "Project documentation scaffolding. Covers about.md, specs.md, architecture.md, project-context.md, and user stories. Use when creating or maintaining project documentation in a .project/ folder, scaffolding specs, architecture docs, or user stories. Keywords: project setup, documentation, specs, architecture, stories."
metadata:
version: "1.0.0"
release_date: "2026-03-11"
---
# Project Creator
Guide creation and maintenance of project documentation in `.project/` folder.
## Quick Navigation
- **Templates**: `assets/` folder
- **Guides**: `references/` folder
## When to Use
- Starting a new project
- Creating/updating `.project/` documentation
- Planning user stories
## Project Structure
```
.project/
├── about.md # Project overview, goals, target audience
├── specs.md # Technical specifications, versions
├── architecture.md # System architecture with diagrams
├── project-context.md # Cumulative decisions (optional)
└── stories/
├── stories.md # Master list
└── 001-story-name.md # Individual stories
```
## Assets (Templates)
| Template | File | Purpose |
| ------------ | ------------------------------------ | ------------------- |
| About | `assets/about-template.md` | Project overview |
| Specs | `assets/specs-template.md` | Technical specs |
| Architecture | `assets/architecture-template.md` | System architecture |
| Context | `assets/project-context-template.md` | Project context |
| Stories List | `assets/stories-template.md` | Stories master list |
| Story | `assets/story-template.md` | Individual story |
## Guides (References)
| Guide | File | Purpose |
| ------------ | ---------------------------------- | -------------------- |
| About | `references/about-guide.md` | Questions & criteria |
| Specs | `references/specs-guide.md` | Questions & criteria |
| Architecture | `references/architecture-guide.md` | Questions & criteria |
| Context | `references/context-guide.md` | Questions & criteria |
| Stories | `references/stories-guide.md` | Questions & criteria |
## Document Dependencies
```
about.md → specs.md → architecture.md → project-context.md → stories/
```
---
## Workflow per Document
### Standard Flow
1. Load guide from `references/`
2. Check if document exists — read and assess if yes
3. Create/update using template from `assets/`
4. Interview user with questions from guide
5. Draft content based on responses
6. Present draft, ask for feedback
7. Iterate until user confirms
8. Move to next document
### Existing Documents
When documents already exist:
- Read and assess completion state
- Ask user: review/update or skip?
- Build on existing content — don't recreate
- Note gaps or inconsistencies
---
## Communication Style
### Always Do
- Ask one topic at a time
- Offer 2-3 formulation variants
- Explain technical concepts simply
- Summarize understanding before drafting
- Confirm before moving to next step
### Never Do
- Assume answers
- Skip confirmation steps
- Use jargon without explanation
- Move to next document without confirmation
- Write specs without verifying latest versions
---
## Example Interaction
```
Agent: Let's start with about.md. Can you describe what this project does in your own words?
User: It's an app for tracking expenses.
Agent: Great start! Here are some formulation options:
1. "A personal finance app for tracking daily expenses"
2. "An expense tracker that helps users manage spending"
3. "A mobile app for recording and categorizing expenses"
Which resonates most? Or combine elements?
User: Option 2, but for small businesses, not personal.
Agent: Updated formulation:
"An expense tracker that helps small businesses
manage spending and categorize business expenses."
Does that capture it? Should we add reporting
or team features?
```
---
## Critical Rules
- **NEVER** assume project details — always ask
- **ALWAYS** offer formulation variants
- **ALWAYS** validate understanding with user
- **NEVER** proceed without user confirmation
- **ALWAYS** use templates from `assets/`
- **NEVER** allow version downgrades (only upgrades)
---
## Example Questions
Questions are **examples** — adapt to each unique project:
### About.md
- Can you describe your project in one sentence?
- What frustration does this solve for users?
- If you could only build 3 features, what would they be?
- How will you know if the project is successful?
- What will this project definitely NOT do?
### Specs.md
- What devices/browsers must be supported?
- How many concurrent users do you expect?
- Is there existing infrastructure to integrate with?
- What's the team's experience with [technology]?
- Are there any compliance requirements?
### Architecture.md
- What are the main parts of the system?
- How should components communicate?
- What data flows exist?
- How will it scale?
- How is it secured?
### Stories
- What's absolutely essential for launch?
- What can wait until later?
- What's the logical order of features?
- Does feature X need feature Y first?
---
## Helping Non-Technical Users
When user lacks technical knowledge:
1. **Explain options simply**
- "React is great for interactive UIs, Vue is simpler to learn"
- "PostgreSQL is reliable for structured data, MongoDB for flexible schemas"
2. **Use analogies**
- "The API is like a waiter taking orders to the kitchen"
- "The database is like a filing cabinet"
- "The cache is like a notepad for quick lookups"
3. **Make recommendations with reasoning**
- "Given your expected user count, I recommend..."
- "Since you need real-time updates, this pattern..."
4. **Explain trade-offs**
- Performance vs. simplicity
- Feature richness vs. learning curve
- Cost vs. scalability
## Links
- [Agent Skills Specification](https://agentskills.io/specification)
© itechmeat · MIT · itechmeat/llm-code