Can I restrict what tools a subagent can use?

Yes, the manage-agents skill allows you to define explicit tool lists, allow all tools, or disable tool access entirely for analysis-only roles.

How do I test a new subagent?

You can invoke agents interactively using the @agent-name syntax in Claude or use the provided Python scripts in the skill repository for programmatic validation.

What is a Claude Code subagent?

A subagent is a specialized version of Claude configured with a specific system prompt, tool permissions, and domain expertise to handle focused tasks within a project.

Does this skill support Model Context Protocol (MCP)?

Yes, you can configure access to specific MCP servers or resources on a per-agent basis, allowing subagents to interact with external data and specialized services.

What is the difference between project and user agents?

Project agents are stored in .claude/agents/ and are typically checked into version control for team use, while user agents are stored in your home directory for personal use across all projects.

Subagent Manager

Name: Subagent Manager
Author: dawiddutoit

bydawiddutoit

0•

Productivity & Workflow

Manages and configures specialized Claude Code subagents with custom capabilities, tool permissions, and domain expertise.

This skill provides a comprehensive framework for creating, modifying, and orchestrating specialized Claude subagents within your development environment. It guides users through defining expert domains—such as Python development or database management—configuring granular tool access, and integrating Model Context Protocol (MCP) servers. By formalizing subagent management, it enables complex, modular workflows where tasks are delegated to specific specialists with curated system prompts, ensuring higher quality output through specialized roles and strict validation gates.

Key Features

01Defined patterns for Orchestrators, Quality Guardians, and Domain Experts

02Granular tool and MCP server permission management

030 GitHub stars

04Support for project-specific (.claude/agents/) and user-wide agents

05Standardized YAML frontmatter for agent metadata and configuration

06Integrated validation and testing scripts for agent behavior

Use Cases

01Creating a 'Quality Guardian' agent for read-only code validation and standards checking

02Setting up a 'Documentation Specialist' with access to specific content management tools

03Configuring an 'Integration Specialist' agent with dedicated MCP server access for external API interactions

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add dawiddutoit/custom-claude manage-agents

For use in Claude.ai and ChatGPT

Download Skill