Mcacp FAQs

Question 1

What is Mcacp and what does it do?

Accepted Answer

Mcacp is an MCP (Model Context Protocol) server that acts as a bridge, connecting any MCP-compatible client (like Claude Code, Zed, VS Code) to various ACP (Agent Client Protocol) coding agents. It transforms a single chat window into a multi-agent control plane for AI-assisted development.

Question 2

How can Mcacp improve my coding and development workflows?

Accepted Answer

Mcacp significantly enhances workflows by letting you run any ACP agent from any MCP client, compare multiple AI agents side-by-side on the same task, build sophisticated multi-agent pipelines, and manage agent permissions interactively. It streamlines complex AI-driven development tasks.

Question 3

Does Mcacp support persistence for agent sessions?

Accepted Answer

Yes, Mcacp allows you to persist and resume agent sessions. You can shut down your environment and later `load_session` to have the agent pick up exactly where it left off, ensuring continuity in your development process.

Question 4

How does permission control work with Mcacp?

Accepted Answer

Mcacp offers flexible permission control. For hosts supporting MCP elicitation (e.g., Zed), requests appear as interactive forms. Otherwise, it defaults to operator mode where the host agent or a defined policy (`allow_all`, `deny_all`, `elicit`, `operator`) decides, providing granular control over agent actions.

Question 5

Can I discover and install new agents through Mcacp?

Accepted Answer

Absolutely. Mcacp allows you to browse the ACP registry directly from your chat client, install new agents, and check for upgrades, all without leaving your conversation environment. This makes managing your AI agent toolkit seamless.

Mcacp

Mcacp

关于

主要功能

使用案例

关于

主要功能

使用案例