Where are my sub-agent configurations stored?

Project-specific agents are stored in '.maestro/agents/', while global user-level agents are stored in your home directory at '~/.maestro/agents/'.

What is a Maestro sub-agent?

A Maestro sub-agent is a specialized AI persona with its own unique system prompt, tool access, and model configuration, designed to handle specific domain tasks like auditing or debugging.

Which models can I use for sub-agents?

Maestro sub-agents support Claude 3.5 Sonnet, Opus, and Haiku models, allowing you to balance performance and cost based on the task's complexity.

Can I restrict which tools a sub-agent can use?

Yes, you can explicitly define a list of allowed tools (such as Read, Grep, or Bash) in the agent's YAML frontmatter to ensure specialized focus and security.

How do I create a new sub-agent?

You can create a sub-agent by adding a markdown file to the '.maestro/agents/' directory or by using the interactive '/agents' command within Claude Code.

Maestro Sub-Agents

Name: Maestro Sub-Agents
Author: scooter-lacroix

byscooter-lacroix

Productivity & Workflow

Orchestrates specialized AI sub-agents with custom toolsets, system prompts, and model configurations for targeted development workflows.

About

Maestro Sub-Agents provides a powerful framework for creating specialized AI personas within Claude Code, allowing developers to delegate complex tasks to isolated contexts. By defining agents through simple markdown files or CLI commands, users can configure specific tool access, choose between different Claude models (Sonnet, Opus, Haiku), and implement custom system prompts. This skill is essential for managing sophisticated workflows like automated code reviews, security auditing, and codebase exploration through agent chaining and resumable sessions.

Key Features

Support for persistent, resumable agent sessions and task chaining
Interactive agent management dashboard via /agents command
Multi-model support for optimizing speed versus reasoning depth
Custom persona definition via markdown configuration files
0 GitHub stars
Granular tool permission management (Read, Write, Bash, etc.)

Use Cases

Creating a low-latency 'Explorer' agent using Claude 3 Haiku for rapid codebase navigation
Establishing a multi-step debugging pipeline where specialized agents identify, fix, and verify bugs
Deploying a dedicated security reviewer agent to audit code changes before merging

About

Key Features

Support for persistent, resumable agent sessions and task chaining
Interactive agent management dashboard via /agents command
Multi-model support for optimizing speed versus reasoning depth
Custom persona definition via markdown configuration files
0 GitHub stars
Granular tool permission management (Read, Write, Bash, etc.)

Use Cases

Creating a low-latency 'Explorer' agent using Claude 3 Haiku for rapid codebase navigation
Establishing a multi-step debugging pipeline where specialized agents identify, fix, and verify bugs
Deploying a dedicated security reviewer agent to audit code changes before merging