FastMCP with Adam Azzam and Jeremiah Lowin

2026-04-07 · Source: Software Engineering Daily · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Software Development & Engineering · Depth: Advanced, extended

Summary

FastMCP, an open-source project stewarded by Prefect, builds upon the Model Context Protocol (MCP) to provide Python developers with ergonomic abstractions for creating and deploying MCP servers and applications. MCP is a crucial standard in agentic AI, enabling LLMs to access external tools and data. Initially a weekend project by Prefect CEO Jeremiah Lowin, FastMCP gained rapid adoption and was integrated into the official MCP SDK. The project has evolved through versions, with FastMCP 3.0 introducing a more opinionated framework approach structured around three pillars: servers, clients, and apps. A significant recent development, "code mode" in FastMCP 3.1, allows LLM clients to submit and execute programs against MCP tools within a secure server-side sandbox, addressing limitations of serial tool calls and context window bloat. This innovation, leveraging projects like Pydantic's Monte, aims to enhance efficiency and flexibility for internal enterprise AI workflows.

Key takeaway

For AI Engineers building agentic applications, FastMCP 3.1's new "code mode" offers a critical advancement. By enabling LLMs to submit and execute programs against your MCP tools within a secure server-side sandbox, you can overcome context bloating and serial tool invocation. This allows for more complex, parallel operations and reduces token consumption, making your internal enterprise AI workflows significantly more efficient and robust. Consider integrating "code mode" to enhance agent capabilities and performance.

Key insights

FastMCP 3.1's "code mode" enables LLMs to execute programs against MCP tools in a secure server-side sandbox, boosting efficiency.

Principles

• Standardized protocols enhance agentic AI interoperability.
• High-level abstractions accelerate developer productivity.
• Server-side execution improves LLM efficiency and security.

Method

FastMCP 3.1's "code mode" allows server authors to expose a search tool and an execute tool. Clients submit code, which is run in a secure sandbox (e.g., Pydantic's Monte) against backend tools.

In practice

• Use FastMCP decorators for easy server tool exposure.
• Implement "code mode" for parallel tool execution.
• Leverage prefab for interactive mini-apps in Python.

Topics

• Model Context Protocol
• Agentic AI
• Large Language Models
• FastMCP
• AI Orchestration
• Code Mode
• Server-side Sandboxing

Best for: CTO, VP of Engineering/Data, Director of AI/ML, AI Engineer, Machine Learning Engineer, AI Architect

Related on AIssential

• AI tools for hearing difficulties — helpful or harmful for language learning? — AI assistive technology presents a dilemma for language learners with hearing difficulties, balancing aid with skill development.
• How Enterprise Marketing Teams Are Actually Using AI Agents in 2026 — Enterprise marketing is shifting from human-assisted AI tools to autonomous AI agents that redesign entire workflows.
• [AINews] Microsoft Build: MAI-Thinking-1 and MAI Family models — Microsoft's MAI models showcase a transparent, full-stack AI strategy emphasizing clean data and local agent execution.
• Are we getting better at explaining things because of our interaction with LLMs? — Crafting clear LLM prompts may enhance human explanatory and communication abilities.
• OpenAI expands Codex with role-specific plugins to build a general-purpose app for non-developers — OpenAI is transforming Codex into a general-purpose, role-specific AI work app, expanding beyond coding for non-developers.
• How to Use Claude Managed Agents? — Anthropic's Claude Managed Agents simplifies AI agent deployment by providing a hosted, stateful, and secure execution environment.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Software Engineering Daily.