> ## Documentation Index
> Fetch the complete documentation index at: https://docs.claude-space.com/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# Skills Node

> Add capabilities to your agent with file operations, API calls, and code execution

# Skills Node

**Skills** give your agent capabilities—tools and actions it can perform.

***

## Overview

**Purpose:** Add tools and capabilities to your agent (file operations, API calls, code execution)
**Connection:** Right handle of main agent card
**Color:** Red
**Icon:** 🛠️
**Required:** Optional (0 to many)

***

## What are Skills?

Skills are folder-based capabilities that define what your agent can do:

* Read and write files
* Execute Python/JavaScript code
* Call external APIs
* Process data
* Interact with systems

***

## Skill Structure

Each skill is a folder containing:

```
my-skill/
├── SKILL.md              # Skill definition (required)
├── scripts/              # Executable scripts
│   ├── main.py
│   └── helper.sh
├── references/           # Documentation
│   └── api-docs.md
└── assets/               # Resources
    └── config.json
```

***

## SKILL.md Format

The skill definition file uses YAML frontmatter + markdown:

```markdown theme={null}
---
name: GitHub API
description: Interact with GitHub repositories
version: 1.0.0
author: Your Name
---

# GitHub API Skill

This skill provides GitHub API integration.

## Usage

Call the skill with:
- Repository name
- Action (get, create, update, delete)
- Parameters (issue number, PR number, etc.)

## Examples

Get repository info:
Input: repo=owner/name, action=get

Create issue:
Input: repo=owner/name, action=create_issue, title="Bug", body="Description"
```

***

## Built-in Skills

ClaudeSpace includes these common skills:

| Skill                | Purpose             | Example Use                  |
| -------------------- | ------------------- | ---------------------------- |
| **File System**      | Read/write files    | Read documents, save reports |
| **Code Interpreter** | Execute Python code | Data analysis, calculations  |
| **HTTP Request**     | Call APIs           | External service integration |
| **Web Search**       | Search the internet | Research, fact-checking      |

***

## Creating Custom Skills

**Via Settings → Node Managers → Skills:**

1. Click "Add New Skill"
2. Enter name and description
3. Write SKILL.md content
4. Add scripts in `/scripts/` folder
5. Add docs in `/references/` folder
6. Save to workspace library

**Skill then appears in left sidebar for dragging to canvas.**

***

## Configuration

**When clicked on canvas:**

* Edit SKILL.md content
* Add/remove script files
* Configure parameters
* Set file permissions

***

## Real-World Examples

### Example 1: Email Sender Skill

**SKILL.md:**

```markdown theme={null}
---
name: Email Sender
description: Send emails via SMTP
version: 1.0.0
---

# Email Sender Skill

Sends emails using configured SMTP server.

## Parameters
- to: Recipient email address
- subject: Email subject line
- body: Email body content
- attachments: Optional file paths

## Usage
Agent can call this skill to send emails on your behalf.
Requires SMTP configuration in settings.
```

**scripts/send\_email.py:**

```python theme={null}
import smtplib
from email.mime.text import MIMEText

def send_email(to, subject, body):
    # SMTP logic here
    pass
```

***

### Example 2: Database Query Skill

**SKILL.md:**

```markdown theme={null}
---
name: PostgreSQL Query
description: Query PostgreSQL database
version: 1.0.0
---

# PostgreSQL Query Skill

Execute SQL queries on configured PostgreSQL database.

## Parameters
- query: SQL query string
- params: Query parameters (for prepared statements)

## Safety
- Read-only queries recommended
- Use prepared statements to prevent SQL injection
```

***

## Best Practices

**✅ Do:**

* Keep skills focused (one responsibility)
* Document parameters clearly
* Include usage examples
* Handle errors gracefully
* Use scripts for complex logic

**❌ Don't:**

* Create monolithic "do everything" skills
* Hardcode credentials (use config)
* Skip error handling
* Ignore security implications

***

## Next Steps

* [Memory Node](/docs/node-types/memory-node) - Add persistent knowledge
* [Commands Node](/docs/node-types/commands-node) - Create slash commands

***

<div align="center">
  [← Back to Node Types](/docs/node-types/overview) | [Agent Node →](/docs/node-types/agent-node) | [Memory Node →](/docs/node-types/memory-node)
</div>