Hey everyone! 👋 We just launched two things that we're pretty excited about: 1️⃣ Prefect MCP Server is out of beta 🎉 You can now ask Claude (or ChatGPT, or Cursor) about your Prefect workspace in plain English. "Why is this flow failing?" → Claude reads your logs and tells you exactly what's wrong "What's running right now?" → Claude checks your deployments and gives you a summary Setup is one command:

claude mcp add prefect -- uvx --from prefect-mcp prefect-mcp-server

2️⃣ Prefect + Pydantic AI Integration 🤝 Build AI agents that resume from failure instead of restarting from scratch. When your agent fails at step 10, it picks up at step 10. Steps 1-9 are cached. You only pay for what needs to retry. Makes AI agents way more reliable (and way cheaper to run). Watch Nate debug a failing pipeline and let Claude approve database cleanups: Read the technical deep dives: - https://www.prefect.io/blog/a-prefect-mcp-server - https://www.prefect.io/blog/prefect-pydantic-integration Try it yourself: 1️⃣ - https://docs.prefect.io/v3/how-to-guides/ai/use-prefect-mcp-server - https://github.com/PrefectHQ/prefect-mcp-server 2️⃣ - https://docs.prefect.io/v3/examples/ai-data-analyst-with-pydantic-ai - https://docs.prefect.io/v3/examples/ai-database-cleanup-with-approval Would love to hear what you all build with these! Drop your questions, feedback, or cool use cases in the thread 👇

I’ve been using it a fair bit, and it’s so helpful!