Mcp Interactive

Enables communication between an LLM and a user through an interactive Electron interface.

An interactive MCP (Model Context Protocol) server with an Electron interface for user interaction.

This MCP server enhances LLM coordination, offering greater control and predictability, and significantly reduces billable requests by minimizing user interactions within a single task.

For a detailed history of changes, see theChangelog.

{ "type": "object", "properties": { "projectName": { "type": "string", "description": "Identifies the context/project making the request" }, "message": { "type": "string", "description": "The specific question for the user. Supports Markdown formatting." }, "predefinedOptions": { "type": "array", "items": { "type": "string" }, "description": "Predefined options for the user to choose from (optional)" } }, "required": [ "projectName", "message" ] }

{ "mcpServers": { "interactive": { "command": "npx", "args": [ "mcp-interactive", "-t", "300" ] } } }

{ "mcpServers": { "interactive": { "command": "node", "args": [ "/path/to/project/mcp-interactive/index.js", "-t", "300" ] } } }

System Prompt Recommendations forVisual Studio Code,Cursor IDE

## Decision Making - Ambiguous tasks: ALWAYS clarify using mcp_interactive_ask_user tool - Requires decisions: ALWAYS clarify using mcp_interactive_ask_user tool ## Code Development - Final confirmation: MUST ask user if all requirements from specification are completed using mcp_interactive_ask_user tool with work report ## Final gate - Before ANY final answer (after actions/edits/checks), it is MANDATORY to call mcp_interactive_ask_user with a report on the work done and the option ["Yes, everything is OK"]. - If the answer is not "Yes, everything is OK" — continue working until the comments are resolved and repeat every time mcp_interactive_ask_user. - There are no exceptions (including minor edits).

Example of "Human In The Loop" mode in chat ofVisual Studio Code

Can you check the docs and explain how the project works? #interactive

System Prompt Recommendations forTrae IDE

This section outlines recommended guidelines for system prompts or rules for LLMs to ensure optimal interaction and task execution.

## Decision Making - Ambiguous tasks: ALWAYS clarify using ask_user tool via MCP - Requires decisions: ALWAYS clarify using ask_user tool via MCP ## Code Development - Final confirmation: MUST ask user if all requirements from specification are completed using ask_user tool via MCP with work report ## Final gate - Before ANY final answer (after actions/edits/checks), it is MANDATORY to call ask_user with a report on the work done and the option ["Yes, everything is OK"]. - If the answer is not "Yes, everything is OK" — continue working until the comments are resolved and repeat every time ask_user. - There are no exceptions (including minor edits).

- --timeoutor-t(seconds): Specifies the waiting time for the user's response dialog. If no response is received within this period, the reply "User did not reply: Timeout occurred." will be sent.

- Ensure the current version of Node.js is installed
- Try reinstalling dependencies by running:npm install

Timeout issues for answer (progressbar):

- Increase timeout value using-tparameter
- Default timeout is 60 seconds, adjust as needed

Contributions are welcome! Please feel free to submit a Pull Request. For major changes, please open an issue first to discuss what you would like to change.

Please make sure to update tests as appropriate and follow the existing code style.

If you encounter any problems or have questions, pleaseopen an issueon GitHub.

This project includes a local copy ofmarked.js(marked.min.js) for Markdown parsing functionality. Marked is licensed under the MIT License and is maintained by the Marked.js team and contributors.

An intelligent, LLM-powered email assistant using the Nylas v3 API.