# Creating Skills

> >-

- **Type:** Skill
- **Install:** `agentstack add skill-outcomeeng-claude-creating-skills`
- **Verified:** Yes — security-reviewed for prompt injection and unsafe behavior
- **Seller:** [outcomeeng](https://agentstack.voostack.com/s/outcomeeng)
- **Installs:** 0
- **Category:** [Agent Skills](https://agentstack.voostack.com/c/agent-skills)
- **Latest version:** 0.1.0
- **License:** MIT
- **Upstream author:** [outcomeeng](https://github.com/outcomeeng)
- **Source:** https://github.com/outcomeeng/claude/tree/main/plugins/claude/skills/creating-skills

## Install

```sh
agentstack add skill-outcomeeng-claude-creating-skills
```

Requires the [AgentStack CLI](https://agentstack.voostack.com/docs/cli). Works with Claude Code, Cursor, and any MCP-compatible agent.

## About

When this skill is invoked, Claude Code provides the base directory in the loading message:

```
Base directory for this skill: /path/to/.claude/plugins/cache/{marketplace}/{plugin}/{version}/skills/creating-skills/
```

Throughout this skill, we refer to this as `${SKILL_DIR}`.

Use this path for all skill file access:

- References: `${SKILL_DIR}/references/`
- Workflows: `${SKILL_DIR}/workflows/`
- Templates: `${SKILL_DIR}/templates/`
- Scripts: `${SKILL_DIR}/scripts/`

**IMPORTANT**: Do NOT search the project directory for skill files. If you cannot find a file, use Glob: `.claude/plugins/cache/**/creating-skills/**/*.md`

Skills are prompts. All prompting best practices apply. Be clear, be direct, assume Claude is smart.

**Pure XML Structure**: No markdown headings (#) in skill body. Use semantic XML tags:

- `` - What the skill does
- `` - Immediate actionable guidance
- `` - How to know it worked

**Progressive Disclosure**: SKILL.md under 500 lines. Details go in `references/` and `workflows/`.

**Router Pattern** (for complex skills):

```
skill-name/
├── SKILL.md              # Router + essential principles
├── workflows/            # Step-by-step procedures (FOLLOW)
├── references/           # Domain knowledge (READ)
├── templates/            # Output structures (COPY + FILL)
└── scripts/              # Executable code (RUN)
```

**Skill Types**: Match structure to purpose:

| Type       | Purpose              | Key Output                   |
| ---------- | -------------------- | ---------------------------- |
| Builder    | Create artifacts     | Code, documents, widgets     |
| Guide      | Provide instructions | Tutorials, workflows         |
| Automation | Execute workflows    | Processed files, deployments |
| Analyzer   | Extract insights     | Reports, summaries           |
| Validator  | Enforce quality      | Pass/fail assessments        |
| Reference  | Share knowledge      | Standards loaded by others   |

**Domain Discovery**: Research the domain BEFORE asking users. Users want expertise IN the skill.

What would you like to do?

1. Create a new skill
2. Audit or improve an existing skill
3. Add a component (workflow, reference, template, script)
4. Understand skill patterns

**Wait for response before proceeding.**

| Response | Workflow |
|----------|----------|
| 1, "create", "new", "build" | `${SKILL_DIR}/workflows/create-new-skill.md` |
| 2, "audit", "improve", "review", "check" | `${SKILL_DIR}/workflows/audit-skill.md` |
| 3, "add workflow" | `${SKILL_DIR}/workflows/add-workflow.md` |
| 3, "add reference" | `${SKILL_DIR}/workflows/add-reference.md` |
| 3, "upgrade to router" | `${SKILL_DIR}/workflows/upgrade-to-router.md` |
| 4, "patterns", "understand", "help" | Read `${SKILL_DIR}/references/skill-patterns.md` |

**Intent-based routing** (if user provides clear context):

- "verify content is current" → `${SKILL_DIR}/workflows/verify-skill.md`
- "audit this skill" → `${SKILL_DIR}/workflows/audit-skill.md`
- "create skill for X" → `${SKILL_DIR}/workflows/create-new-skill.md`

**After reading the workflow, follow it exactly.**

**YAML Frontmatter** (required):

```yaml
---
name: skill-name # lowercase-with-hyphens, ≤64 chars
description: >- # Directive, ≤1024 chars. Add NEVER only if it disambiguates.
  ALWAYS invoke this skill when .
---
```

**Simple Skill Structure**:

```text
What the skill does
Minimal working example
Step-by-step procedure
How to know it worked
```

**Router Skill Structure**:

```text
Always applies
Question to ask user
Maps answers to workflows
Available references
Available workflows
```

**Naming Convention**: Prefer gerund form (verb + -ing):

- `creating-skills`, `processing-pdfs`, `reviewing-code`

All in `${SKILL_DIR}/references/`:

| File                    | Purpose                                          |
| ----------------------- | ------------------------------------------------ |
| core-principles.md      | XML structure, conciseness, degrees of freedom   |
| use-xml-tags.md         | Required and conditional XML tags                |
| skill-patterns.md       | Type-specific patterns, templates, assets        |
| reusability-patterns.md | Variations vs constants, adaptable skills        |
| testing-patterns.md     | Evaluation-driven development, iterative testing |
| technical-patterns.md   | Error handling, security, dependencies           |

All in `${SKILL_DIR}/workflows/`:

| Workflow             | Purpose                                |
| -------------------- | -------------------------------------- |
| create-new-skill.md  | Build a skill from scratch             |
| audit-skill.md       | Check skill against best practices     |
| add-workflow.md      | Add a workflow to existing skill       |
| add-reference.md     | Add a reference to existing skill      |
| upgrade-to-router.md | Convert simple skill to router pattern |
| verify-skill.md      | Check if content is still accurate     |

All in `${SKILL_DIR}/templates/`:

| Template            | Purpose                       |
| ------------------- | ----------------------------- |
| simple-skill.md     | Single-file skill scaffold    |
| router-skill.md     | Router pattern skill scaffold |
| builder-skill.md    | Builder type template         |
| guide-skill.md      | Guide type template           |
| automation-skill.md | Automation type template      |
| analyzer-skill.md   | Analyzer type template        |
| validator-skill.md  | Validator type template       |

All in `${SKILL_DIR}/scripts/`:

| Script            | Purpose                              |
| ----------------- | ------------------------------------ |
| init_skill.py     | Initialize skill directory structure |
| package_skill.py  | Validate and package skill           |
| quick_validate.py | Quick YAML/structure validation      |

A well-structured skill:

- Has valid YAML frontmatter (name + description)
- Uses pure XML structure (no markdown headings in body)
- Has essential principles inline in SKILL.md (if router pattern)
- Routes to appropriate workflows based on user intent
- Keeps SKILL.md under 500 lines
- Has been tested with real usage

## Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

- **Author:** [outcomeeng](https://github.com/outcomeeng)
- **Source:** [outcomeeng/claude](https://github.com/outcomeeng/claude)
- **License:** MIT

Install and usage instructions live in the source repository linked above.

## Pricing

- **Free** — Free

## Security capabilities

Automated source analysis of v0.1.0 — what this tool can access:

- **Network access:** no
- **Filesystem access:** no
- **Shell / process execution:** no
- **Environment & secrets:** no
- **Dynamic code execution:** no

*"Yes" means the capability is present in the source — more access means more to trust, not that it is unsafe.*


## Versions

- **0.1.0** — security scan: passed — Imported from the upstream source.

## Links

- Listing page: https://agentstack.voostack.com/l/skill-outcomeeng-claude-creating-skills
- Seller: https://agentstack.voostack.com/s/outcomeeng
- Browse the marketplace: https://agentstack.voostack.com/browse

---
Listed on AgentStack — the marketplace for AI agent skills and MCP servers. Every listing is security-reviewed. Creators keep 70%.