.NET Aspire 13.2 Ships an Agent-Native CLI — and It Changes How You Build Distributed AI Apps

TL;DR: Aspire 13.2 dropped on March 23, 2026, and its headline feature isn’t just a new integration — it’s a rethought CLI purpose-built for AI coding agents. If you’re shipping agentic .NET workloads on Azure, this release gives your coding agent the same operational superpowers you’ve always had in the dashboard.

What Shipped (and When)

Aspire 13.2 landed on March 23rd, 2026 — the same day as Aspire Conf. That’s not a coincidence. The team wanted you to be able to watch a session and immediately run the thing. Smart. The headline features are TypeScript AppHost authoring, an AI-agent-native CLI, new and improved integrations, and a smarter dashboard. Each of those would be a decent blog post on its own, but the agent-native CLI is the one that matters most right now for teams shipping agentic workloads.

The Core Problem It Solves

Coding agents need structure to build on and visibility into what’s happening. Aspire 13.2 focuses on giving agents the same power Aspire gives you, so they can do complex work without you copy-pasting logs, manually restarting processes, and sending screenshots. In other words: your AI coding agent was previously flying blind inside a multi-service Aspire app. It could write code, but it couldn’t observe running resources, restart a misbehaving service, or wait for health checks. You had to babysit it. That ends here.

The New Agent-Friendly CLI Commands

Agents can now start your AppHost in the background with a new --detach mode, stop and start any resource from the CLI without tearing down the whole system, and wait for resource statuses to update before moving on to their next action. The new command surface looks like this:

# Start the AppHost in the background so the agent keeps going
aspire start --detach

# Restart just the API without nuking everything else
aspire resource api restart

# Block until the API is healthy (up to 2 minutes)
aspire wait api --status healthy --timeout 120

The --isolated flag lets agents spin up parallel environments without stepping on each other — random ports, separate secrets, and no more conflicts across git worktrees. This is a big deal for teams running multiple Copilot agents or worktree-based PR workflows simultaneously. The aspire docs command brings aspire.dev documentation directly into agent workflows so agents can search and retrieve docs programmatically without needing additional MCPs configured. One less MCP server to wire up — your agent can self-serve the docs. Plus, aspire doctor validates your whole environment before an agent starts building. Think of it as a pre-flight check that saves you from the “agent burned 40 tokens before realizing Node.js wasn’t installed” experience we all know and love.

TypeScript AppHost: Polyglot Gets Real

Aspire 13.2 continues deep investment in multi-language support, allowing you to write your AppHost completely in TypeScript. For teams with a significant TypeScript/Node presence alongside their .NET services, this removes the last excuse for not adopting Aspire as the orchestrator. Modern applications rarely live in a single language or framework, and Aspire is embracing that reality head-on. Aspire 13.2 continues expanding support for polyglot, multi-language applications, making it easier to model, run, observe, and deploy systems made up of many different technologies. Whether your services are written in C#, Python, JavaScript, TypeScript, or a mix of all of them, Aspire provides shared primitives and a consistent experience across the whole system. —

The Wider Ecosystem Context: Foundry + MCP

Aspire 13.2 doesn’t exist in isolation. In 13.2, the team added support for Certbot, overhauled the AI Foundry integration for the next-gen Microsoft Foundry, brought Bun into the JavaScript integration, and more. The overhauled Foundry integration matters because Microsoft Foundry brings together model access, orchestration, evaluation, and deployment into a single environment, and is used to develop intelligent solutions such as copilots and automated workflows — connecting AI to real-world tools and services, and moving projects from experimentation to secure, production-ready systems on Azure. Meanwhile, on the DevOps side, the Remote MCP Server is a hosted version of the Azure DevOps MCP Server that uses streamable HTTP transport; it supports the same core scenarios as the local server but removes the need for additional setup and installation. Pair that with Aspire 13.2’s agent CLI and you have a coding agent that can query your backlog, restart a failing service, and wait for the health check — all without leaving its loop. That’s agentic DevOps in practice, not in theory.

Practical Integration Steps

Here’s a minimal path to get your Copilot agent running against an Aspire 13.2 app:

1. Install or update Aspire:

dotnet tool update -g aspirate
# or via the Aspire CLI
aspire update --self

2. Initialize with agent support (from Aspire 13.1+):

aspire init
# Follow the prompts to configure your agent environment
# (VS Code, GitHub Copilot CLI, Claude Code, etc.)

3. Wire up the Remote Azure DevOps MCP Server in mcp.json:

{
  "servers": {
    "ado-remote-mcp": {
      "url": "https://mcp.dev.azure.com/{organization}",
      "type": "http",
      "headers": {
        "X-MCP-Toolsets": "repos,wiki,wit",
        "X-MCP-Readonly": "true"
      }
    }
  }
}

⚠️ Preview caveat: Support for Azure AI Foundry, Microsoft 365 Copilot, and Copilot Studio in the Remote ADO MCP Server is not yet available but will be added soon. For now, Visual Studio and VS Code are your supported clients.

4. Start detached and let the agent iterate:

aspire start --detach
aspire doctor  # sanity check before the agent touches anything

Cost & Latency Considerations

The Azure DevOps MCP Server itself is free to use. However, standard Azure DevOps pricing applies to your organization and any data access through the service. AI assistant usage may have separate costs depending on your chosen AI platform. For the Foundry side, Managed long-term memory in Foundry Agent Service is in public preview and is free during preview — you pay only for the underlying model calls. Budget accordingly: the agent loop can be chatty, so pick a cost-optimized model for the “is this service healthy?” calls and reserve the heavy reasoning models for code generation.

What to Watch Next

Aspire 13.2 full What’s New — the announcement teases breaking changes worth reviewing before you upgrade production pipelines.
Remote ADO MCP → Foundry support — once that lands, you’ll be able to build Foundry agents that query your backlog natively without a local process.

AI agents are already improving the .NET MAUI contribution workflow, reportedly reducing issue resolution time by 50–70% while increasing test coverage and code quality — a signal that agentic DevOps on .NET is leaving the prototype stage.