Marionette FAQs

Question 1

Does Marionette only work with GitHub Copilot?

Accepted Answer

No, Marionette implements the open Model Context Protocol (MCP) and is designed to work with any AI client that supports MCP servers. This includes GitHub Copilot in VS Code, Claude Desktop, Claude Code, and Cursor.

Question 2

What is Marionette and its primary purpose?

Accepted Answer

Marionette is a server that implements the Model Context Protocol (MCP) to bridge AI agents with humans. Its primary purpose is to enable human-in-the-loop AI workflows for decision-making, clarification, and dynamic session steering via structured forms.

Question 3

What are the benefits of using Marionette with AI assistants like GitHub Copilot?

Accepted Answer

Marionette offers significant cost savings for AI tools with premium request models, like GitHub Copilot. By providing structured human input and early session steering, it drastically reduces unnecessary back-and-forth and wasted requests, potentially cutting costs by 50-70% on complex tasks.

Question 4

What kind of scenarios is Marionette best suited for?

Accepted Answer

Marionette excels in scenarios requiring human oversight or input, such as architecture decisions, clarifying ambiguous requirements, confirming risky refactors, managing long-running tasks, and enabling human checkpoints in multi-step workflows.

Question 5

How does Marionette facilitate human-AI interaction?

Accepted Answer

It provides tools like `askQuestion` for structured queries with predefined options and `steerSession` for humans to redirect the AI's goal. This allows agents to pause execution and wait for specific human input, making interactions precise and efficient.

Marionette

Marionette

主な機能

ユースケース

主な機能

ユースケース