Can I restrict the tools available to a specific subagent?

Yes, you can use the optional 'tools' field in the YAML frontmatter to specify a whitelist of tools, such as ["Read", "Grep", "Glob"], to limit the agent's capabilities.

What permission modes are available for subagents?

Subagents support several permission modes including default, acceptEdits, dontAsk, bypassPermissions, and plan, which control how the agent handles sensitive actions.

What is the recommended length for a subagent system prompt?

System prompts should be specific, concrete, and must be kept under 10,000 characters to maintain efficiency and performance.

Where should subagent files be stored in a project?

Subagent files should be placed in either the .claude/agents/ or .claude-plugin/agents/ directory to be recognized by the system.

What are the required frontmatter fields for a subagent?

Every subagent requires a name (3-50 characters in kebab-case), a description with 2-4 usage examples, a model setting (inherit, haiku, sonnet, or opus), and a UI color.

Subagent Implementation

Name: Subagent Implementation
Author: v1truv1us

byv1truv1us

Productivity & Workflow

Standardizes the creation and configuration of specialized subagents within the Claude Code environment.

About

This skill provides a comprehensive technical reference for developers looking to build, configure, and deploy custom subagents in Claude Code. It offers structured guidance on YAML frontmatter specifications, system prompt engineering patterns, and file organizational standards, ensuring that custom agents are triggered correctly and perform consistently across different tasks. By following these implementation patterns, users can extend Claude's capabilities with domain-specific logic, restricted toolsets, and optimized model selection for specialized workflows.

Key Features

Standardized YAML frontmatter templates for agent metadata
0 GitHub stars
Best practices for defining agent tools, permissions, and model selection
Validated file naming conventions and directory placement logic
Quick checklists for verifying subagent syntax and operational behavior
System prompt structure guidelines for role definition and task processing

Use Cases

Creating domain-specific documentation assistants within a project
Building specialized code reviewers or unit test generators
Defining custom agents with restricted tool access for security-sensitive tasks

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add v1truv1us/plugin-marketplace subagent-implementation

For use in Claude.ai and ChatGPT

Download Skill

GitHub