Phone-a-Friend MCP Server 🧠📞

Name: Phone-a-Friend MCP Server 🧠📞
Author: abhishekbhakat

@abhishekbhakat

代码仓库

关于 Phone-a-Friend MCP Server 🧠📞

暂无概览

#phone-a-friend #ai-consultation #problem-solving

基本信息

分类

其他

许可证

MIT license

运行时

python

传输方式

stdio

发布者

abhishekbhakat

配置

使用下面的配置,将此服务器添加到你的 MCP 客户端。

{
  "mcpServers": {
    "phone-a-friend-mcp-server": {
      "command": "uv",
      "args": [
        "pip",
        "install",
        "-e",
        "."
      ]
    }
  }
}

工具

未检测到工具

工具是从 README 中自动提取的。维护者可以在 ## Tools 标题下列出工具,即可填充这部分内容。

概览

What is Phone-a-Friend MCP Server?

Phone-a-Friend MCP Server is an AI-to-AI consultation system that enables one AI to “phone a friend” (another AI) for critical thinking, long context reasoning, and complex problem solving via OpenRouter. It creates a two-step process: packaging context for an external AI to analyze, then extracting actionable insights for the primary AI. It’s designed for AI systems that need to leverage other AI models as consultants for difficult tasks.

How to use Phone-a-Friend MCP Server?

Install by cloning the repository and running uv pip install -e .. Set one or more API keys via environment variables (OPENROUTER_API_KEY recommended, or OPENAI_API_KEY, ANTHROPIC_API_KEY, GOOGLE_API_KEY). Start the server with phone-a-friend-mcp-server. To use with Claude Desktop, add a configuration entry to claude_desktop_config.json. The primary AI invokes the phone_a_friend tool with all_related_context, task, and optional any_additional_context.

Key features of Phone-a-Friend MCP Server

Two-step consultation: context + reasoning then extract actionable insights.
Automatic model selection based on provider and task.
Supports OpenRouter, OpenAI, Anthropic, and Google/Gemini.
Exposes a single phone_a_friend MCP tool.
Handles long context reasoning (>100k tokens).
No memory of past conversations; quality depends on provided context.

Use cases of Phone-a-Friend MCP Server

Complex multi‑step problems requiring deep analysis.
Situations needing long context reasoning with extensive information.
Cross‑domain expertise consultation.
Critical decision‑making with high stakes.
Problems that benefit from multiple AI perspectives.

FAQ from Phone-a-Friend MCP Server

How does Phone-a-Friend MCP Server differ from a simple chained model call?

It packages all relevant context and reasoning instructions into a structured two‑step workflow, then extracts insights into a usable format for the primary AI. The external AI has no memory of previous calls, so the quality depends entirely on the completeness of the context provided.

What runtime dependencies does Phone-a-Friend MCP Server require?

Python with uv and pip. All Python dependencies are installed via uv pip install -e .. API keys must be set as environment variables. No additional databases or external services are required besides the chosen AI provider.

Where does user data go when using Phone-a-Friend MCP Server?

All context is sent to the configured AI provider (OpenRouter, OpenAI, Anthropic, or Google) for analysis. The server itself does not persist any data; processing is done in‑memory per request.

What are the known limitations of Phone-a-Friend MCP Server?

It is not intended for simple factual questions, basic reasoning tasks, quick responses, or well‑defined procedural tasks. The external AI has no memory of prior conversations, so every call starts fresh. The quality of the reasoning depends heavily on the quality and completeness of the context provided.

What transports and authentication does Phone-a-Friend MCP Server support?

The server uses the Model Context Protocol (MCP) – likely stdio transport when run locally. Authentication is handled via API keys passed as environment variables (OPENROUTER_API_KEY, OPENAI_API_KEY, etc.). No built‑in HTTP or authentication layers beyond that.