Installs into .claude/skills of the current project.
Are you the author of Add Skill?
Add the live security badge to your README. It updates with every re-scan.
[](https://www.skillsdirectory.com/skills/diegosouzapw-add-skill)
---
name: add-skill
description: Guide for adding a new skill to an existing plugin. Use when creating or adding skills to plugins in this marketplace.
---
# Add Skill Guide
Guide for adding a new skill to an existing plugin in this marketplace.
## Skill Directory Structure
```
plugins/<plugin-name>/skills/<skill-name>/
├── SKILL.md # Required: skill definition
├── *.md # Optional: additional instructions
└── scripts/ # Optional: executable scripts
└── *.sh
```
## SKILL.md Structure
Every skill requires a `SKILL.md` file with YAML frontmatter:
```markdown
---
name: skill-name
description: Brief description of what this skill does and when to use it
---
# Skill Name
## Instructions
[Clear, step-by-step guidance for Claude to follow]
## Examples
[Concrete examples of using this skill]
```
## Required Fields
### name
- Maximum 64 characters
- Lowercase letters, numbers, and hyphens only
- Cannot contain "anthropic" or "claude"
### description
- Must be non-empty
- Maximum 1024 characters
- Include both what the skill does AND when to use it
## Three Levels of Content
### Level 1: Metadata (always loaded)
YAML frontmatter with `name` and `description`. Loaded at startup (~100 tokens).
### Level 2: Instructions (loaded when triggered)
Main body of SKILL.md. Loaded when skill is triggered (under 5k tokens
recommended).
### Level 3: Resources (loaded as needed)
Additional files and scripts. Only loaded when referenced.
## Using Scripts
For complex logic, create executable scripts:
```
skills/<skill-name>/scripts/
└── do-something.sh
```
Benefits:
- Script code never enters context window
- Only script output consumes tokens
- More reliable than generated code
## Script Path Resolution
**Important**: Skills are loaded from a plugin cache, not the repository.
Hardcoded paths like `plugins/foo/skills/bar/scripts/script.sh` will NOT work.
When referencing scripts in SKILL.md, use this pattern:
```markdown
**Script location**: `scripts/my-script.sh` (relative to this skill's directory)
Before running, locate this skill's directory (where this SKILL.md is located),
then execute: bash <skill-directory>/scripts/my-script.sh
```
This ensures Claude finds the correct script path regardless of where the plugin
is installed.
## Reference
- Agent Skills overview:
https://platform.claude.com/docs/en/agents-and-tools/agent-skills/overview
- Best practices:
https://platform.claude.com/docs/en/agents-and-tools/agent-skills/best-practices