aep-executor

Manage background task and agent lifecycles across Claude Code, Codex, and tmux+cmux backends.

Updated May 18, 2026

Search Tags:#Claude Code #worktree #agent management #executor #background task

Ready to run

npx skills add https://github.com/memorysaver/project-understanding --skill aep-executor

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: aep-executor
Source: https://github.com/memorysaver/project-understanding/tree/main/.agents/skills/aep-executor
Command: npx skills add https://github.com/memorysaver/project-understanding --skill aep-executor

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires aep, and includes scripts (resource) and references (resource) components.

What problem does it solve?

This Skill addresses the challenge of managing and executing background tasks across different agents and workflows in a standardized, efficient manner.

Core Features & Use Cases

Unified Executor Abstraction: Defines a standardized interface for launching, steering, and managing background agents.
Cross-Backend Compatibility: Works seamlessly with Claude Code, Codex, and tmux+cmux backends.
Human-Gate Integration: Allows mid-task decision points to be routed to a human for review and approval.
Automated Worktree Management: Manages the creation and teardown of worktrees for each task execution.

Quick Start

To start a new background agent with the aep-executor skill, run:

/aep-launch -w .feature-workspaces/myworkspace

Frequently Asked Questions about aep-executor

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

FAQPage Schema

How do I manage background tasks across multiple agents in Claude Code?▼

You can manage background tasks across agents using a unified executor abstraction that launches, steers, and manages background agents in Claude Code. It standardizes the lifecycle of task execution across different workflows.

Can I run background agents in Codex and tmux+cmux using the same executor?▼

Yes, the executor supports cross-backend compatibility, allowing you to run and manage background agents seamlessly across Claude Code, Codex, and tmux+cmux environments using a standardized interface.

How do I handle human approval gates during automated task execution?▼

Human-gate integration allows mid-task decision points to be routed to a human for review and approval. This ensures automated task execution can pause for human oversight before proceeding.

Do I need to manually set up worktrees for background agent execution?▼

No, automated worktree management handles the creation and teardown of worktrees for each task execution automatically. This streamlines background agent workflows without manual workspace setup.

What is the best way to recover from errors in background agent workflows?▼

Robust error handling mechanisms facilitate error recovery in background agent workflows. The executor manages the task lifecycle to ensure automated execution can recover from failures across various backends.

How do I launch a new background agent in a specific workspace?▼

To launch a new background agent, run the `/aep-launch` command with the `-w` flag followed by your workspace path. This initiates task execution in the specified worktree.