Is the sub-agent's work visible in my main chat?

Sub-agents operate in the background. You must use the READ action or wait for the agent's final report to see the results of their work.

When should I use an agent instead of a standard tool?

Use an agent for complex, multi-step explorations that require independent reasoning, or when you need to perform multiple distinct research tasks at the same time.

How many agents can I run at once?

You can spawn multiple agents with unique numeric IDs (e.g., agent 0, 1, 2) to handle different research or analysis tasks in parallel.

Can I limit what tools a sub-agent can use?

Yes, you can use the allowed_tools and disallowed_tools parameters to strictly control the sub-agent's capabilities, such as making them read-only.

What is a Claude sub-agent?

A sub-agent is an independent Claude session spawned to perform specific tasks autonomously, reporting its findings back to the main orchestrator once completed.

Claude Multi-Agent Orchestrator

Name: Claude Multi-Agent Orchestrator
Author: cyrup-ai

bycyrup-ai

生产力与工作流

Spawns and manages autonomous sub-agents to perform parallel research and complex delegated tasks within Claude Code.

关于

This skill enables Claude to act as an orchestrator by spawning independent sub-sessions that can execute tasks autonomously. It allows for parallel processing of complex research, independent code analysis, and multi-step delegations where each sub-agent operates in its own context with specific tool permissions. By delegating background work to these agents, developers can handle multiple streams of inquiry or analysis simultaneously without cluttering the main conversation, significantly increasing productivity during large-scale refactors or deep-dive audits.

主要功能

Granular tool control with allowlist and blocklist permissions
Autonomous deep-dive research and codebase analysis capabilities
Parallel sub-agent spawning for concurrent task execution
0 GitHub stars
Custom system prompts and conversation turn limits for sub-sessions
Real-time status tracking and management of active agent sessions

使用场景

Conducting parallel research on multiple error patterns across a large codebase simultaneously
Performing concurrent API documentation and test coverage analysis across different directories
Delegating a security audit to an autonomous sub-agent while continuing feature development in the main session

关于

主要功能

Granular tool control with allowlist and blocklist permissions
Autonomous deep-dive research and codebase analysis capabilities
Parallel sub-agent spawning for concurrent task execution
0 GitHub stars
Custom system prompts and conversation turn limits for sub-sessions
Real-time status tracking and management of active agent sessions

使用场景

Conducting parallel research on multiple error patterns across a large codebase simultaneously
Performing concurrent API documentation and test coverage analysis across different directories
Delegating a security audit to an autonomous sub-agent while continuing feature development in the main session