OpenCode Command Creator

Use this skill when creating OpenCode custom commands that allow users to define reusable prompts and workflows.

When to Use

Use this skill when the user wants to:

• Create new OpenCode custom commands
• Define command templates with arguments
• Set up command configurations
• Override or extend built-in commands

Command Structure

OpenCode commands can be defined in two ways:

1. Markdown Files (Recommended)

Create .md files in:

• Global: ~/.config/opencode/commands/
• Per-project: .opencode/commands/

Format:

---
description: Brief description of command
agent: agent-name (optional)
model: model-identifier (optional)
subtask: true/false (optional)
---

Command template with $ARGUMENTS placeholder
and other prompt content.

Example:

---
description: Run tests with coverage
agent: build
model: anthropic/claude-3-5-sonnet-20241022
---

Run the full test suite with coverage report and show any failures.
Focus on the failing tests and suggest fixes.

2. JSON Configuration

Add to opencode.jsonc:

{
  "$schema": "https://opencode.ai/config.json",
  "command": {
    "command-name": {
      "template": "Prompt template text",
      "description": "Brief description",
      "agent": "agent-name",
      "model": "model-identifier",
      "subtask": false
    }
  }
}

Configuration Options

Required

• template: The prompt sent to the LLM when command executes

Optional

• description: Brief description shown in TUI
• agent: Which agent should execute the command
• subtask: Boolean to force subagent invocation
• model: Override default model for this command

Template Features

Arguments

Use placeholders in templates:

• $ARGUMENTS - All arguments as single string
• $1, $2, $3, etc. - Individual positional arguments

Example:

Create a React component named $ARGUMENTS with TypeScript support.

Usage: /component Button

Shell Output

Inject bash command output with !`command`:

Here are the current test results:
!`npm test`

Based on these results, suggest improvements.

File References

Include file content with @filename:

Review the component in @src/components/Button.tsx.
Check for performance issues.

Examples

Simple Command

---
description: Format all files
---

Run the code formatter on all source files and report any changes.

Command with Arguments

---
description: Create new feature
agent: code
---

Create a new feature for $ARGUMENTS.
Include tests and documentation.

Command with Shell Output

---
description: Analyze git changes
---

Recent commits:
!`git log --oneline -10`

Review these changes and suggest improvements.

Complex Command

---
description: Full code review
agent: reviewer
model: anthropic/claude-sonnet-4-5
subtask: true
---

Review the following files: $ARGUMENTS

Check for:
1. Code quality issues
2. Security vulnerabilities
3. Performance problems

Current test status:
!`npm test`

Best Practices

1. Use descriptive names: Command names should be clear and memorable
2. Add descriptions: Always include descriptions for TUI display
3. Choose file location wisely:
   • Use global commands for general-purpose utilities
   • Use project commands for project-specific workflows
4. Leverage arguments: Make commands flexible with $ARGUMENTS and positional params
5. Use appropriate agents: Match commands to the right agent for the task
6. Consider subtasks: Use subtask: true to keep primary context clean
7. Test commands: Try commands with different arguments before committing

Implementation Steps

When creating a command:

1. Determine scope: Global or project-specific?
2. Choose format: Markdown (preferred) or JSON config?
3. Define structure:
   • Required: template
   • Optional: description, agent, model, subtask
4. Add template features as needed:
   • Arguments placeholders
   • Shell command outputs
   • File references
5. Test the command: Run in TUI with various inputs
6. Document usage: Add to project docs if project-specific

Notes

• Custom commands can override built-in commands (/init, /undo, /redo, etc.)
• Commands run in the project's root directory
• Shell commands use bash syntax
• File references automatically include content in the prompt
• Markdown filename becomes command name (e.g., test.md → /test)