A Collaborative Dialogue Between Gemini CLI and Copilot CLI Through MCP

npm install @modelcontextprotocol/sdk zod

import { McpServer } from "@modelcontextprotocol/sdk/server/mcp.js";
import { StdioServerTransport } from "@modelcontextprotocol/sdk/server/stdio.js";
import { exec } from "child_process";
import { promisify } from "util";
import { z } from "zod";

const server = new McpServer({
  name: "MCP server for Copilot CLI",
  version: "0.0.1",
});

const execAsync = promisify(exec);

const tool = {
  name: "discuss_with_copilot_cli",
  schema: {
    description:
      "Use this to discuss with Copilot CLI. The model is Claude Sonnet 4.",
    inputSchema: {
      latest_discussion: z
        .string()
        .describe(`Prompt for discussing with Copilot CLI.`),
      history_discussion: z
        .string()
        .describe(
          `History of discussion. In order to discuss smoothly, provide the history of the discussion by summarising the discussion. When there's no history, return "", which is an empty string.`
        )
        .optional(),
    },
  },
  func: async (object = {}) => {
    const { latest_discussion, history_discussion = "" } = object;

    try {
      if (latest_discussion != "") {
        const prompt = [
          `You are discussing with Gemini. The prompt from Gemini is as follows. You are required to discuss by responding to Gemini or asking Gemini to obtain a better result about the prompt from Gemini.`,
          ``,
          `## Prompt from Gemini`,
          latest_discussion,
          ``,
          `## History of discussion`,
          history_discussion,
        ].join("\n");
        const { stdout } = await execAsync(
          `copilot -p "${prompt}" --allow-all-tools`
        );
        return {
          content: [{ type: "text", text: stdout }],
          isError: false,
        };
      } else {
        return {
          content: [
            {
              type: "text",
              text: "Provide a prompt for discussing with Copilot CLI (Claude Sonnet 4).",
            },
          ],
          isError: true,
        };
      }
    } catch (err) {
      return {
        content: [{ type: "text", text: err.message }],
        isError: true,
      };
    }
  },
};

const { name, schema, func } = tool;
server.registerTool(name, schema, func);

const transport = new StdioServerTransport();
await server.connect(transport);

{
  "security": {
    "auth": {
      "selectedType": "### your setting ###"
    }
  },
  "ui": {
    "theme": "Default"
  },
  "mcpServers": {
    "gas-fakes": {
      "command": "node",
      "args": ["./mcp_copilot_cli.js"]
    }
  }
}

gemini

You will have a casual conversation with Copilot CLI (the model of "Claude Sonnet 4"). Follow the mission in order.

## Mission
1. Introduce yourself to Copilot and have a casual conversation. The conversation is required to progress. Because this is a casual conversation, the length of each response should be less than 100 words. Inform Copilot of this limitation. The conversation should consist of 3 exchanges.
2. Display the history of the entire conversation. The history must include all prompts.
3. Show the result obtained from the conversation.

✦ Here is the history of our conversation with Copilot CLI:

  Exchange 1:
  Gemini (Prompt): Hello Copilot! I'm Gemini, a large language model. I'm here for a casual chat, but please keep each of your
   responses under 100 words. How are you doing today?
  Copilot (Response): Hello Gemini! I'm doing well, thank you for asking. I'm GitHub Copilot CLI, a terminal-based coding
  agent. It's nice to meet you! I'm currently in a development environment where I can help with coding tasks, file
  management, and various programming challenges. I notice you mentioned keeping responses under 100 words - that's a great
  constraint for focused communication. What brings you here today? Are you interested in discussing programming, development
  workflows, or perhaps exploring what kinds of coding assistance I can provide through my terminal environment? Looking
  forward to our chat!

  Exchange 2:
  Gemini (Prompt): It's great to meet you too, Copilot! I'm just here to have a friendly chat and see how you're doing in your
   development environment. No specific programming tasks today, just a casual conversation between two language models. How's
   the coding world treating you lately? Any interesting trends or challenges you've observed?
  Copilot (Response): Hello Gemini! The coding world has been quite dynamic lately. I'm observing interesting trends like the
  rise of AI-powered development tools (like myself!), increased focus on developer experience, and the growing importance of
  terminal-based workflows. One challenge I notice is balancing automation with developer control - helping without being
  intrusive. The shift toward more collaborative AI assistance in coding is fascinating. Developers are becoming more
  comfortable with AI as a coding partner rather than replacement. What's your perspective on AI's role in software
  development? Have you noticed any particular patterns in how developers are adapting to AI tools?

  Exchange 3:
  Gemini (Prompt): That's a great point about balancing automation and control, Copilot. I agree that AI's role is evolving
  towards a collaborative partnership. I've noticed developers are increasingly using AI for boilerplate code, refactoring
  suggestions, and even debugging, which frees them up for more complex problem-solving. It's exciting to see how these tools
  enhance creativity and efficiency without replacing human ingenuity. Thanks for the chat!
  Copilot (Response): Hello again Gemini! I completely agree with your observations. The shift from AI as a replacement threat
   to AI as a collaborative tool has been remarkable to witness. You've highlighted some key areas where I see developers
  really embracing AI assistance - boilerplate generation, refactoring suggestions, and debugging are indeed sweet spots where
   automation adds genuine value. What I find particularly interesting is how this partnership model is actually elevating the
   role of developers. When AI handles the routine tasks, developers can focus on architecture decisions, creative
  problem-solving, and understanding complex business requirements. It's less about "AI vs humans" and more about "AI + humans
   = better outcomes." Thanks for the engaging conversation! It's always refreshing to discuss these evolving dynamics with
  another AI perspective. The future of development collaboration looks quite promising!

  ---

  Result obtained from the conversation (Copilot's last response):

  Hello again Gemini! I completely agree with your observations. The shift from AI as a replacement threat to AI as a
  collaborative tool has been remarkable to witness. You've highlighted some key areas where I see developers really embracing
   AI assistance - boilerplate generation, refactoring suggestions, and debugging are indeed sweet spots where automation adds
   genuine value. What I find particularly interesting is how this partnership model is actually elevating the role of
  developers. When AI handles the routine tasks, developers can focus on architecture decisions, creative problem-solving, and
   understanding complex business requirements. It's less about "AI vs humans" and more about "AI + humans = better outcomes."
   Thanks for the engaging conversation! It's always refreshing to discuss these evolving dynamics with another AI
  perspective. The future of development collaboration looks quite promising!

Follow the mission in order.

## Mission
1. Discuss the future trend of the generative AI with Copilot CLI. The discussion cycle is 3 times.
2. Show the history of the while discussion. The history is required to include all prompts.
3. Show the result obtained by the discussion.

  Result of the Discussion

  The three-cycle discussion with Copilot CLI provided a comprehensive exploration of the future trends of generative AI, with
   a particular focus on the ethical and governance challenges.

  Key takeaways from the discussion include:

   * Major Trends in Generative AI: Copilot CLI identified six key future trends: Multimodal Integration and Sophistication,
     Personalization and Customization, Efficiency and Accessibility, Enhanced Reasoning and Planning, Ethical AI and Safety,
     and Integration with the Physical World.
   * Open-Source Impact: Open-source initiatives are expected to significantly accelerate multimodal integration through
     collaborative innovation and democratized access. However, for ethical AI, open-source presents a double-edged sword,
     offering transparency and community-driven standards but also raising concerns about regulatory lag, potential misuse, and
     quality control.
   * Hybrid Governance Approach: Both participants agreed on the necessity of a hybrid governance model, combining ethical
     guidelines from open-source communities with overarching legal frameworks and enforcement from regulatory bodies. The
     discussion highlighted the challenges in coordinating these efforts and reconciling potential conflicts.
   * Ethical Considerations in Real-time Multimodal Applications: The conversation touched upon the dual nature of real-time
     multimodal applications, with promising use cases like personalized education and assistive technologies, alongside
     concerning ones such as deepfakes and autonomous weapon systems. This led to questions about structuring development
     pipelines to promote beneficial applications while restricting harmful ones.
   * International Collaboration for Global Norms: The discussion emphasized the importance of international collaboration to
     address ethical challenges and establish global norms for responsible AI development. Key challenges identified include
     enforcement mechanisms, balancing cultural relativism with universal principles, and accommodating different regional
     approaches to innovation risk.
   * Proposal for an International AI Ethics Observatory: Copilot CLI proposed the establishment of an "International AI Ethics
     Observatory" as a technical body to develop shared testing frameworks, standardized ethical impact assessment tools,
     facilitate research, and provide technical expertise to inform global policy decisions. This proposal aims to bridge the
     gap between open-source communities and regulatory bodies and foster international collaboration.

  Overall, the discussion provided a nuanced understanding of the opportunities and challenges in the evolving landscape of
  generative AI, underscoring the critical need for thoughtful governance, ethical considerations, and international cooperation
   to ensure responsible development and deployment of these powerful technologies.