acp

Delegate software development tasks to external ACP-compatible coding agents.

Updated Aug 27, 2026
One-click install
npx skills add https://github.com/CerisonAutomation/clud --skill acp-cerisonautomation
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: acp
Source: https://github.com/CerisonAutomation/clud/tree/main/assistant/src/config/bundled-skills/acp
Command: npx skills add https://github.com/CerisonAutomation/clud --skill acp-cerisonautomation

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

It solves the need to delegate complex coding work to specialized external agents that can edit files and run independently, while streaming results back to the user.

Core Features & Use Cases

  • Spawns external coding agents via ACP: Runs dedicated adapters such as Claude Code and Codex as subprocesses over stdio, streaming output into the conversation.
  • Supports concurrent agent sessions: Enables parallel work to handle multiple coding tasks at once.
  • Session control and interruption: Lets you list running agents, check status, abort a session, or steer a running agent with new instructions.

Quick Start

Ask the assistant to use acp_spawn to delegate your coding task to the claude agent in the current project folder.

Frequently Asked Questions about acp

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

FAQPage Schema
How do I delegate coding tasks to autonomous agents for concurrent execution?

To delegate coding tasks to autonomous agents, you use the acp_spawn command to run external ACP-compatible coding adapters like claude-agent-acp or codex-acp as subprocesses over stdio, streaming results directly back into your conversation. This enables concurrent agent sessions for parallel work.

What is ACP and how does it work for workspace orchestration?

ACP is a protocol that delegates complex software development tasks to external coding agents for autonomous execution. It works by spawning adapter binaries as subprocesses over stdio, allowing you to implement features, debug, and refactor code within a chosen working directory while streaming results back.

Do I need specific adapter binaries installed to use ACP for debugging and refactoring?

Yes, you need an ACP-enabled environment with properly installed adapter binaries such as claude-agent-acp or codex-acp. These adapters are spawned via acp_spawn to execute debugging and refactoring tasks autonomously, requiring correct session management and safe working directory control via cwd.

Can I steer or interrupt a running coding agent during autonomous execution?

Yes, you can steer or interrupt a running coding agent during autonomous execution. ACP supports session control and interruption, allowing you to list running agents, check their status, abort a session entirely, or steer a running agent with new instructions as it works.

How do I manage the working directory when running multiple coding agents concurrently?

You manage the working directory when running multiple coding agents concurrently by using the cwd parameter for safe working directory control. This ensures that spawned external ACP-compatible agents operate within the correct project folder boundaries during autonomous execution.

What are the limitations of using ACP to delegate complex software development?

Limitations of using ACP to delegate software development include the strict requirement for an ACP-enabled environment with correctly installed adapter binaries. Additionally, safe working directory control via cwd is mandatory to prevent autonomous agents from editing files outside the chosen project folder.