subagent-task

Decompose complex work into specialized parallel sub-tasks via the task tool.

3|1|Updated Apr 12, 2026
One-click install
npx skills add https://github.com/Cogni-AI-OU/cogni-ai-agent-skills --skill subagent-task
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: subagent-task
Source: https://github.com/Cogni-AI-OU/cogni-ai-agent-skills/tree/main/subagent-task
Command: npx skills add https://github.com/Cogni-AI-OU/cogni-ai-agent-skills --skill subagent-task

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

It helps you decompose complex, multi-step work into specialized sub-tasks so your main agent can coordinate efficiently and avoid context overload.

Core Features & Use Cases

  • Delegation governance: Enforces rules for when and how to spawn sub-agents, including allowed targets and prohibiting unapproved built-in subagents.
  • Parallelizable workflows: Supports concurrent sub-agent execution for independent activities such as fact retrieval, log analysis, and plan validation.
  • Synthesis & safety checks: Requires the coordinator to synthesize each sub-agent’s output and handle unexpected replies or failures with clear user-facing reporting.
  • Use Case: When preparing a PR fix that needs code review, security audit, and test validation, delegate each dimension to the appropriate project agent(s) and merge results into a single actionable response.

Quick Start

Use the subagent-task skill when you have a complex or parallelizable workflow and you need the task tool to invoke specialized project agents while keeping the main agent as the coordinating synthesizer.

Frequently Asked Questions about subagent-task

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I coordinate sub-agents for parallel codebase analysis and task execution?

To coordinate sub-agents for parallel execution, use the task tool to decompose complex work into specialized sub-tasks. This allows your main agent to manage concurrent activities like codebase analysis and log review while avoiding context overload.

What is the best way to decompose multi-step agent workflows into specialized sub-tasks?

The best way to decompose multi-step agent workflows is by delegating independent activities to specialized project agents via the task tool. The main agent then synthesizes each sub-agent's output and handles unexpected replies or failures with clear user-facing reporting.

Can I run parallel sub-agents for security checks and plan validation without context overload?

Yes, you can run parallel sub-agents for independent activities like security checks and plan validation. By delegating these dimensions to appropriate project agents, the main coordinator agent remains free of context overload and merges results into a single actionable response.

How does delegation governance work when spawning sub-agents via the task tool?

Delegation governance enforces rules for when and how to spawn sub-agents, including specifying allowed delegation targets and prohibiting unapproved built-in subagents. The coordinator maintains strict responsibilities for prompt clarity and task_result synthesis.

What should I do when a spawned sub-agent fails or returns unexpected replies during complex operations?

When a spawned sub-agent fails or returns unexpected replies during complex operations, the coordinator agent must handle the error through clear user-facing reporting. The coordinator is strictly responsible for synthesizing outputs and reporting failures safely.

When should I not use sub-agents for agent operations and multi-step work?

You should not use sub-agents for multi-step work if the tasks cannot be clearly decomposed into independent parallel activities, or if you lack defined allowed delegation targets. Proper governance and prompt clarity are required to prevent unapproved subagent usage.