# Create Claude Code Agent Skills

Teams need repeatable ways to package workflows and domain knowledge for AI agents. This skill guides Claude Code through designing, writing, and testing focused Agent Skills.

## Install

```bash
npx skillstore add egadams/create-new-skills
```

## Metadata

- Status: approved
- Slug: egadams-create-new-skills
- Skillstore revision: r1
- Version status: missing
- Tree hash: 55d4074b5ae5187da29095ae99e7e9cb6a95dd3f7c126173fd849aafa8b55775
- Author: EGAdams
- GitHub username: EGAdams
- License: MIT
- Repository: https://github.com/EGAdams/planner/tree/main/.claude/skills/meta-skill
- Ref: 1ffa7643651792ccb4bd3b15d924d2c97edff755
- Supported tools: Claude, Codex, Claude Code
- Audit status: complete
- Agent install advisory: allowed
- Manual install advisory: allowed
- Artifact signature: available
- Audit attestation: unavailable
- Human verification: not\_verified
- Risk factors: filesystem, external\_commands
- Quality score: 78
- Quality tier: bronze
- Public page: https://skillstore.pages.dev/skills/egadams-create-new-skills
- Manifest: https://skillstore.pages.dev/api/skills/egadams-create-new-skills/manifest

## Capabilities

- Explains how Agent Skills are structured with SKILL.md frontmatter and Markdown instructions.
- Guides users through defining a skill purpose, triggers, and workflow before writing files.
- Provides examples for simple skills, script-backed skills, documentation skills, and skill extensions.
- Recommends project skill placement for team sharing and personal skill placement for individual use.
- Covers testing steps for file structure, frontmatter, and trigger behavior.

## Use Cases

- Package a team workflow: Turn a repeatable engineering workflow into a shared project skill with clear triggers and instructions.
- Create a personal productivity skill: Capture an individual workflow as a personal Claude Code skill with examples and supporting references.
- Design a script-backed skill: Plan a skill that includes helper scripts, templates, or supporting documentation using progressive disclosure.

## Prompt Templates

### Create a basic skill

```
Create a Claude Code skill for reviewing Python code style. Ask me any missing questions before writing it.
```

### Turn a workflow into a skill

```
Create a project skill from this workflow: first inspect changed files, then run tests, then summarize release risk.
```

### Add supporting files

```
Create a documentation skill that uses a style guide, an API template, and examples as supporting files.
```

### Extend an existing skill

```
Update our existing transcribe skill so it applies project-specific spelling corrections from a separate reference file.
```

## Limitations

- It provides authoring guidance but does not guarantee a skill will trigger in every conversation.
- It relies on the user to supply the workflow, domain rules, and validation examples.
- It includes documentation examples for shell commands that should be reviewed before use.
- It is focused on Claude Code skills, not every skill format across all Claude surfaces.

## Best Practices

- Define the skill trigger and user goal before writing files.
- Keep the main SKILL.md focused and move detailed references into supporting files.
- Test the skill with realistic user requests after creating or changing it.

## Anti Patterns

- Do not combine unrelated workflows into one broad skill.
- Do not use vague descriptions that make skill activation unreliable.
- Do not add scripts or shell commands unless the user needs executable automation.

## Security Audit

- Audited at: 2026-07-09T04:00:55.112\+00:00
- Summary: All static findings were adjudicated as false positives caused by documentation examples, Markdown code formatting, and legitimate Claude Code skill path references. No prompt injection, data exfiltration intent, unauthorized command execution, or hidden malicious behavior was found in the reviewed files.

## Stats

- Views: 176
- Downloads: 7
- Favorites: 0
- Popularity score: 0