Tools, Resources and Prompts

2026-01-03 · Source: Daily Dose of Data Science · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Software Development & Engineering · Depth: Intermediate, medium

Summary

The MCP (Model-Client-Protocol) framework defines three core capabilities for AI servers: Tools, Resources, and Prompts. Tools are executable functions, often with side effects or external API calls, that an AI model can invoke, such as a `get_weather` function. These are model-controlled but typically require user permission for execution. Resources provide read-only data, like databases or knowledge bases, which the AI can query for information without side effects, such as reading a file via `file:///user/notes.txt`. Resources are usually accessed under host application control and are less dangerous but still require privacy and permission considerations. Prompts are predefined templates or conversation flows supplied by the server to guide AI behavior, like a "code reviewer" system role. These are user or developer-controlled, setting the stage before the model generates a response, and allow for server-side updates without client app changes.

Key takeaway

For AI Engineers designing robust AI applications, understanding the MCP framework's distinction between Tools, Resources, and Prompts is crucial. You should implement external actions as Tools with explicit user consent, provide read-only context via Resources, and manage conversational flows with server-side Prompts. This approach enhances control, safety, and maintainability, allowing you to update AI behaviors without modifying client applications.

Key insights

MCP defines Tools, Resources, and Prompts as core capabilities for structured AI interaction and control.

Principles

• AI actions require human oversight.
• Separate read-only data from executable functions.
• Externalize prompt logic for flexibility.

Method

The MCP framework structures AI server capabilities into Tools (model-invoked actions), Resources (host-controlled data retrieval), and Prompts (user/developer-controlled conversational templates) to manage AI interactions.

In practice

• Implement `get_weather` as an MCP tool.
• Use `file:///` URIs for MCP resources.
• Define a "code reviewer" prompt template.

Topics

• Model-Controlled Protocol
• AI Tools
• AI Resources
• AI Prompts
• LLM Function Calling

Best for: AI Engineer, Machine Learning Engineer, Software Engineer

Related on AIssential

• Skills, Tools and MCPs - What’s The Difference? — AI agent capabilities evolve from basic function calls to standardized protocols and explicit expertise layers.
• Prompt Engineering: Saying what matters most — Effective prompts are critical for guiding LLMs, defining behavior, controlling output quality, and managing operational costs.
• What is MCP? — MCP provides a universal interface for AI models to interact with external tools and real-time data.
• From Tool Calling to MCP: Building a Natural Language Search with Spring AI and MCP Server — MCP standardizes AI model communication with external tools, enabling reusability and independent deployment.
• AI Agents, Tools, MCP, and Skills: The Core, The Embellishment, and The Gimmick — AI agent development requires distinguishing core components like Agents and Tools from less practical elements like MCP.
• How to correctly use MCP servers with your AI Agents — Careful MCP server integration prevents context bloat and optimizes AI agent performance and cost.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Daily Dose of Data Science.