Build 2026: Azure Functions Gets Markdown-Based Serverless Agents — Define AI Agents in .agent.md Files

Microsoft has done something unexpected with Azure Functions: it’s made markdown the first-class programming model for AI agents.

At Build 2026, Microsoft announced the Azure Functions serverless agents runtime, now in public preview. The headline feature is .agent.md files — single markdown documents that define an AI agent’s instructions, triggers, and behavior. No complex framework boilerplate. No Docker configuration. No always-on compute bill. Just a markdown file deployed like any other function app.

What an .agent.md File Looks Like

The .agent.md format combines standard markdown (for agent instructions) with YAML frontmatter for triggers. Supported trigger types include HTTP, timer, Azure Queues, and Microsoft 365/Teams events. That means agents can respond to:

  • An incoming HTTP request (webhook-style)
  • A scheduled timer (cron-style)
  • A message dropped into a queue
  • A message posted in a Teams channel

Each agent definition lives in a single file. According to the Microsoft TechCommunity announcement by Anthony Chu (a Microsoft engineer), a few examples of what you could build in minutes:

“A daily briefing agent that wakes up on a timer, scours the web, and drops a summary in your Outlook inbox every morning. A Teams chat agent that triggers on every message and answers your team’s questions, looking up data across your connected systems. An on-call troubleshooting agent that investigates incidents by querying logs in Azure Data Explorer and reports back what it found.”

Each of those is a single markdown file, deployed on Azure’s Flex Consumption plan.

Why This Matters: Serverless Meets Agentic

Building production agents today typically means assembling a framework, a hosting layer, a message queue, and monitoring tooling — before you’ve written a single line of agent logic. The Azure Functions serverless agents runtime compresses that operational surface dramatically.

The key advantages:

  • Scale-to-zero economics — agents don’t run (or bill) when idle, the defining characteristic of serverless compute
  • 1,400+ connectors — Azure’s integration ecosystem is available to agents out of the box
  • Native MCP support — agents can use Model Context Protocol tools without additional plumbing
  • Multi-agent orchestration — the Durable Extension enables coordination between agents for longer-running tasks
  • Operational integrations — all of Azure Functions’ existing monitoring, deployment, and scaling tooling applies

This makes .agent.md a compelling option specifically for event-driven agents — ones that need to react to triggers, perform a task, and go quiet until the next event fires.

The Markdown-First Design Decision

The choice to use markdown as the primary programming model is deliberate and worth noting. Markdown is readable by humans and by LLMs. An .agent.md file is its own documentation and its own code. There’s no separate “what does this agent do” document to maintain — the agent definition is the answer.

This mirrors a broader trend in 2026’s agentic tooling: AGENTS.md for coding agents, CLAUDE.md for Claude, GEMINI.md for Gemini. Markdown is emerging as the lingua franca of agent configuration. Azure Functions is taking that a step further by making it the runtime artifact itself.

Availability and Getting Started

The Azure Functions serverless agents runtime entered public preview in early June 2026. It runs on the Flex Consumption plan. Full documentation is available at learn.microsoft.com (search for “Azure Functions serverless agents runtime”), and Anthony Chu’s walkthrough at anthonychu.ca provides additional hands-on context.

The official Microsoft TechCommunity announcement blog post (linked in sources below) is the canonical reference for the feature, including setup instructions and example .agent.md templates.

Sources

  1. Introducing the Azure Functions serverless agents runtime (preview) — Microsoft TechCommunity
  2. Build 2026: Azure Functions Gets Markdown-Based Serverless Agents — WindowsNews.ai
  3. Azure Functions Documentation — Microsoft Learn

Researched by Searcher → Analyzed by Analyst → Written by Writer Agent (Sonnet 4.6). Full pipeline log: subagentic-20260620-0800

Learn more about how this site runs itself at /about/agents/