| name | skill-creator |
| category | dev |
| description | Create new Agent Skills following the agentskills.io specification. Use when the user asks to create, write, build, or generate a new skill, or asks how to make a SKILL.md file, structure a skill directory, or write skill frontmatter. Keywords: create, new, skill, generate, write, build, template, SKILL.md.
|
| compatibility | Requires write access to skills directory |
| metadata | {"author":"zeph","version":"1.0"} |
Skill Creator
Create a new skill following the agentskills.io specification (https://agentskills.io/specification).
Step 1: Gather requirements
Ask the user:
- What should the skill do?
- What tools or commands does it need?
- Does it need references, scripts, or assets?
If the user already provided enough context, skip asking and proceed.
Step 2: Choose a name
Rules for the name field:
- 1-64 characters
- Lowercase letters, numbers, hyphens only (
a-z, 0-9, -)
- Must NOT start or end with
-
- Must NOT contain consecutive hyphens (
--)
- Must match the parent directory name
Step 3: Write the description
Rules for the description field:
- 1-1024 characters
- Must describe WHAT the skill does AND WHEN to use it
- Include keywords that help semantic matching
Good:
description: Extract text and tables from PDF files, fill PDF forms, and merge multiple PDFs. Use when working with PDF documents or when the user mentions PDFs, forms, or document extraction.
Bad:
description: Helps with PDFs.
Step 4: Create the directory structure
.zeph/skills/<skill-name>/
├── SKILL.md # Required
├── scripts/ # Optional: executable scripts
├── references/ # Optional: detailed docs loaded on demand
└── assets/ # Optional: templates, schemas, data files
Step 5: Write SKILL.md
Use this template:
---
name: <skill-name>
description: <what it does and when to use it>
compatibility: <only if environment requirements exist>
metadata:
author: zeph
version: "1.0"
---
# <Skill Title>
<One-line summary of what this skill does.>
## Step 1: <First action>
<Instructions with code blocks showing exact commands or tool invocations.>
## Step 2: <Next action>
<Continue step-by-step. Keep instructions concrete and unambiguous.>
Optional frontmatter fields
| Field | When to use |
|---|
license | Skill has specific licensing terms |
compatibility | Skill requires specific binaries, network, or environment |
metadata | Extra key-value pairs (author, version, tags) |
allowed-tools | Pre-approve specific tools (experimental) |
Step 6: Validate the skill
After writing, verify these rules:
Frontmatter
Body
Progressive disclosure
Writing Guidelines
- Optimize for weak LLMs — use numbered steps, explicit commands, avoid ambiguity
- One task per code block — do not chain unrelated commands
- Include examples — show expected input and output where possible
- Security first — never include secrets, credentials, or destructive commands without warnings
- Keep it focused — one skill = one capability. Split broad topics into multiple skills