subagent-fanout

Parallelize independent sub-jobs across fresh-context subagents and aggregate results.

746|130|Updated Jun 30, 2026
One-click install
npx skills add https://github.com/Archive228/loopkit --skill subagent-fanout
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: subagent-fanout
Source: https://github.com/Archive228/loopkit/tree/main/skills/subagent-fanout
Command: npx skills add https://github.com/Archive228/loopkit --skill subagent-fanout

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) components.

What problem does it solve?

This Skill helps to avoid context rot by parallelizing independent sub-jobs across fresh-context subagents, enabling efficient handling of tasks that branch into many independent pieces.

Core Features & Use Cases

  • Parallel Task Execution: Spawn one subagent per independent unit (e.g., one file, one source, one check), each with a fresh context.
  • Result Synthesis: An orchestrator synthesizes the results from these subagents without doing the per-unit work itself.
  • Tight Role Definition: Assigns each worker a specific role with only the necessary input.
  • Appropriate Use Cases: Ideal for breadth-first tasks like research, multi-file edits, and multi-source verification. Avoid using when there are sequential dependencies that require step N-1's output.

Quick Start

Run the 'subagent-fanout' skill to process multiple independent items in parallel, e.g., 'subagent-fanout process_items --items item1 item2 item3'.

Frequently Asked Questions about subagent-fanout

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

FAQPage Schema
How do I prevent context rot when processing multiple independent files in parallel?

To prevent context rot during parallel processing, spawn fresh-context subagents for each independent unit so they execute without prior conversation history. An orchestrator subagent then synthesizes the isolated results to maintain processing efficiency.

What is the best way to distribute independent sub-jobs across multiple subagents?

Distributing independent sub-jobs across subagents involves assigning each worker a specific role with only necessary input data. This tight role definition ensures parallel tasks execute efficiently without redundant context loading.

Can I use parallel subagents for tasks with sequential dependencies?

Parallel subagents cannot be used for tasks with sequential dependencies that require step N-1's output. This approach is only appropriate for genuine independent pieces that do not require sequential execution.

How do I run multi-source verification using fresh-context subagents?

Running multi-source verification with fresh-context subagents requires spawning one subagent per source to check data independently. The orchestrator subagent then aggregates these parallel verification results.

When should I avoid using a subagent fanout pattern for task distribution?

You should avoid using a subagent fanout pattern for task distribution when your workflow contains sequential dependencies. If step N requires the output of step N-1, parallel execution will fail to aggregate results properly.