Commands Node

Create custom slash commands for specialized agent behaviors

Commands Node

Commands create custom slash commands for quick agent actions.

---

Overview

Purpose: Define user-facing slash commands (e.g., /summarize, /review-pr) Connection: Configured via panel (no visual connection) Color: Purple Icon: ⚙️ Required: Optional (0 to many)

---

What are Commands?

Commands are shortcuts that trigger specific agent behaviors with parameters:

Example:

User types: /summarize report.pdf
Agent knows: Read file, create 3-bullet summary

---

Command Structure

Commands are defined as markdown files with YAML frontmatter:

---
name: summarize
description: Summarize a document
parameters:
  - name: file
    description: File path to summarize
    required: true
---

# Summarize Command

When user runs `/summarize <file>`:
1. Read the file at the given path
2. Extract key points
3. Return 3-bullet summary

## Output Format
📄 **Summary of [filename]:**
• [Key point 1]
• [Key point 2]
• [Key point 3]

---

Creating Commands

Via Settings → Node Managers → Commands:

1. Click “Add New Command”
2. Define command name (e.g., review-pr)
3. Write YAML frontmatter (parameters, description)
4. Write markdown body (instructions)
5. Save to workspace library

Usage in chat:

/review-pr 123

---

Command Examples

Example 1: Document Summarizer

---
name: summarize
description: Create bullet-point summary
parameters:
  - name: file
    required: true
    description: Document to summarize
  - name: bullets
    required: false
    default: 3
    description: Number of bullet points
---

# Summarize Command

Read the file and create a {bullets}-point summary.
Format as markdown list.

Usage:

/summarize report.pdf
/summarize report.pdf bullets=5

---

Example 2: Code Reviewer

---
name: review-pr
description: Review GitHub pull request
parameters:
  - name: pr_number
    required: true
    description: PR number to review
  - name: focus
    required: false
    default: all
    description: Focus area (security, performance, style, all)
---

# Review PR Command

1. Fetch PR #{pr_number} from GitHub
2. Review code changes
3. Focus on {focus} issues
4. Report findings with severity levels

Usage:

/review-pr 456
/review-pr 456 focus=security

---

Example 3: Deployment Runner

---
name: deploy
description: Deploy application to environment
parameters:
  - name: env
    required: true
    description: Environment (staging, production)
  - name: confirm
    required: false
    default: false
    description: Skip confirmation prompt
---

# Deploy Command

Deploy the application to {env} environment.

## Steps
1. Run tests
2. Build application
3. Deploy to {env}
4. Verify deployment

⚠️ If env=production and confirm=false, ask for confirmation first.

Usage:

/deploy staging
/deploy production confirm=true

---

Parameter Types

Required Parameters

parameters:
  - name: file
    required: true
    description: File to process

Agent will error if not provided.

---

Optional Parameters with Defaults

parameters:
  - name: format
    required: false
    default: markdown
    description: Output format

Agent uses default if not provided.

---

Multiple Parameters

parameters:
  - name: source
    required: true
  - name: target
    required: true
  - name: options
    required: false

Usage: /command source=a target=b options=c

---

Best Practices

✅ Do:

• Use descriptive command names
• Document all parameters clearly
• Provide defaults for optional params
• Include usage examples
• Define expected output format

❌ Don’t:

• Create commands for one-time tasks
• Use generic names (/do, /run)
• Skip parameter descriptions
• Make too many required params

---

Commands vs. Skills

Feature	Commands	Skills
User-facing	Yes	No
Parameters	User provides	Agent decides
Purpose	Shortcuts	Capabilities
Example	/summarize	File System skill

Use both: Command triggers agent, which uses skills to execute.

---

Next Steps

• MCPs Node - External integrations
• Hooks Node - Lifecycle events

---

← Memory Node | Agent Node | MCPs Node →