max-swarm

Orchestrate parallel Codex subagents to execute multi-file tasks.

Updated Aug 23, 2026
One-click install
npx skills add https://github.com/ahostbr/dotcodex --skill max-swarm
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: max-swarm
Source: https://github.com/ahostbr/dotcodex/tree/main/skills/max-swarm
Command: npx skills add https://github.com/ahostbr/dotcodex --skill max-swarm

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Spawn multiple Codex subagents in parallel to work on a task. Only use this skill when the user has explicitly asked for a swarm, parallel agents, or delegated multi-agent work.

Core Features & Use Cases

  • Decompose tasks into 2-5 independent subtasks that can be executed in parallel.
  • Spawn a Codex subagent for each subtask with clear ownership and non-reversion of others' edits.
  • Provide a final synthesis and status report once agents complete.

Quick Start

Decompose the user's task and spawn a swarm of Codex agents to execute the subtasks concurrently.

Frequently Asked Questions about max-swarm

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

FAQPage Schema
How do I run parallel coding agents on a multi-file project?

You can run parallel coding agents by spawning a swarm of Codex subagents. This orchestrates 2-5 independent subtasks concurrently across your codebase, assigning each agent clear ownership to execute multi-file work simultaneously.

When should I use parallel agents for codebase tasks?

Use parallel agents for codebase tasks when you have explicitly requested delegated multi-agent work or a swarm. It is designed for complex multi-file projects that can be decomposed into 2-5 independent subtasks for concurrent execution.

How do I decompose a codebase task for parallel execution by agents?

To decompose a codebase task for parallel execution, break the project into 2-5 independent subtasks. Each subtask is assigned to a spawned Codex subagent with clear ownership, ensuring agents do not revert each other's edits during concurrent execution.

How are parallel subtask results synthesized after agent execution?

Parallel subtask results are synthesized through a final status report. Once the coordinated workflow of spawned agents completes, the system provides a final synthesis that clearly reports the results and overall execution status.

Can I spawn more than five agents for a single project task?

No, the swarm coordination approach decomposes tasks into 2-5 independent subtasks. This limit ensures clear ownership and effective parallel execution across the codebase without overwhelming the coordinated workflow.