Fly.io Engineering Blog & Tech Articles
Global application platform enabling developers to deploy and scale applications with edge computing capabilities
215 engineering articles, tutorials, and technical insights from Fly.io's engineering team
Top Technologies
Latest Articles
Filter:
To start, here’s the status quo: every Fly Machine gets a private IPv6 address, and every other Machine in your organization can reach it. We call that network 6PN. It’s what my-db.internal points at, and it’s how your app server talks to your databa
Daniel Botha, Peter Cai
12 min read
Includes Code
--
CVE-2026-99999
The rest of this won’t make much sense without a clear read on the bug itself, so: the Linux kernel CVE team and pwn.ai announced CVE-2026-99999 on the morning of the 15th. We heard about it that night, and started working on mitigation
Daniel Botha
8 min read
Includes Code
--
OpenClaw needs generous permissions to be useful. Run it in a hardware-isolated Sprite instead of on a Mac Mini.
We’re seeing this play out all across the infra landscape, including here at Fly.io, where agents have overtaken humans as our most common customer species. Thing is, we think it’s important to remember that there’s a human with aspirations behind th
Daniel Botha
6 min read
Includes Code
--
When work on Sprites started, it was a new kind of infra for a new kind of product. So when we couldn’t find the right filesystem in the pasta section, we made our own. It was clever, and that’s exactly why we didn’t like it. This post introduces its
Daniel Botha
7 min read
Includes Code
--
Sprites are a place to run agents; the first thing you should think to do with a new Sprite is to type claude (or gemini or codex). We’ve put a lot of effort into making sure coding agents feel safe and happy when they’re on Sprites, because, to (pro
This is a complicated post. So I need you to promise me something: if you read past this introduction, you’ll read the whole rest of the way through. It’s an honor thing.
A couple months back, Theo Browne ran a video rating the “best place to host a
Kurt Mackey
13 min read
Includes Code
--
Building agents is fun. Rebuilding agents that break themselves... less so. A lot of Fly people are teaching their agents to do anything risky in a Sprite, so a wrong move trashes a sandbox instead of the host. Here's how.
Daniel Botha
8 min read
Includes Code
--
You don't need a Mac Mini to run your AI company. You need a Sprite.
Daniel Botha
6 min read
Includes Code
--
A 2.4 billion parameter language model, quantized to 1.58 bits, running on a Sprite at 50 tokens per second, serving its own Python client to any Sprite that asks.
Daniel Botha
6 min read
Includes Code
--
Sprites are a place to run agents; the first thing you should think to do with a new Sprite is to type claude (or gemini or codex). We’ve put a lot of effort into making sure coding agents feel safe and happy when they’re on Sprites, because, to (pro
Kurt Mackey
6 min read
Includes Code
--
Context engineering is the biggest lever you have over AI agent quality — here's what it looks like when you invest in it over a few weeks of real work.
Daniel Botha
6 min read
Includes Code
--
Give every agent its own computer, its own network policy, and the ability to roll back mistakes without taking down the swarm.
Daniel Botha
4 min read
Includes Code
--
An AI that can see the results of its own actions behaves fundamentally differently than one generating code into a void.
Stop context-switching. Give each project its own environment, its own AI context, and its own checkpoint history.
AI generates code faster than you can review it. You need somewhere safe to run it.
This article by Ben Johnson explains two major new features added to Litestream's Virtual File System (VFS): writable mode and background hydration.
Sprites can spin up and control other Sprites, which turns out to be useful for a surprising number of things.
Slack bots are handy, but they typically require OAuth, which means you need a server with a public URL to handle callbacks. Sprites make this trivial.
Daniel Botha
3 min read
Includes Code
--
Code from your phone without managing servers, VPNs, or complex setups. Sprites and sprite-mobile make mobile coding surprisingly simple.
This article explains the design and implementation of Sprites, Fly. io's new product offering instant-creation Linux VMs with 100GB durable storage backed by object storage.
Fly. io introduces 'Sprites' — durable, persistent cloud computers designed as an alternative to ephemeral sandboxes for AI coding agents.
Litestream VFS is a new SQLite plugin that enables querying databases directly from S3-compatible object storage without downloading the entire database.
Thomas Ptacek argues that every developer should build an LLM agent to truly understand the technology, demonstrating through progressive Python code examples that a functional agent with tool use ...
The article discusses a phishing incident involving Fly. io's CEO, Kurt Mackey, detailing how the attack was executed and the lessons learned from it.
Thomas Ptacek
7 min read
Includes Code
Has Summary
--
The article discusses MorphLLM, an innovative solution for AI agent builders that addresses the inefficiencies of traditional code editing methods.
Daniel Botha
4 min read
Includes Code
Has Summary
--
The article discusses trust calibration, a crucial concept for AI software builders that involves aligning user trust with product capabilities.
Daniel Botha
10 min read
Has Summary
--
The article discusses the innovative approach of using games as a method for evaluating AI models, specifically through the deployment of AI Town on Fly. io.
Daniel Botha
4 min read
Has Summary
--
The article discusses the limitations of model agnosticism in AI projects, emphasizing that understanding and specializing in a specific model is more beneficial than trying to maintain flexibility...
Daniel Botha
5 min read
Has Summary
--
Phoenix. new is a fully-online coding agent designed for Elixir and Phoenix, enabling rapid development of collaborative, real-time applications.
MCP servers, part of the Model Context Protocol, provide a standardized way to connect AI models to various data sources and tools, akin to a USB-C port for AI applications.
Sam Ruby
7 min read
Has Summary
--
The article discusses the skepticism surrounding AI-assisted programming, particularly focusing on the capabilities of Large Language Models (LLMs) in software development.
The article discusses the deployment of Kamal 2. 0 in production environments, highlighting the challenges faced and the solutions available for effective software deployment.
The article discusses a complex concurrency bug encountered in Fly. io's Anycast router, implemented in Rust.
Litestream is an open-source tool designed to enhance the reliability of SQLite applications by enabling them to recover data from object storage.
The article discusses the launch of Model Context Protocol (MCP) servers on Fly. io, highlighting their utility, configuration, and security considerations.
The article discusses the use of Model Context Protocols (MCPs) for provisioning machines, highlighting their integration with tools for managing resources in applications.
The article discusses the implementation of a basic Model Context Protocol (MCP) server for the flyctl command-line tool, highlighting its utility in connecting large language models (LLMs) to vari...
Kurt Mackey
5 min read
Includes Code
Has Summary
--
The article discusses the shift in user demographics on Fly. io's platform, highlighting how robots have become the primary drivers of growth.
Kurt Mackey
10 min read
Includes Code
Has Summary
--
The article discusses the operationalization of Macaroon tokens at Fly. io, detailing their implementation, benefits, and challenges.
The article discusses a performance issue encountered with the Rust-based proxy service, fly-proxy, which is part of Fly. io's infrastructure.
The article discusses the challenges and realizations Fly. io faced while integrating GPU support into their cloud services.
The article features an exit interview with JP Phillips, a key member of the Fly. io engineering team, who reflects on his four years at the company.
The article discusses Fly. io's approach to blogging, emphasizing a shift away from highly curated content towards more spontaneous and frequent posts.
Thomas Ptacek
4 min read
Has Summary
--
The article discusses the evolving capabilities of Semgrep, a semantics-aware code search tool, and its integration with AI-driven development agents like Cursor.
The article discusses the integration of VSCode with SSH for remote editing, particularly in the context of LLM-generated code.
Thomas Ptacek
3 min read
Has Summary
--
The article discusses the integration of Livebook, FLAME, and the Nx stack to create AI GPU clusters that can be operated from a laptop.
The article discusses the concept of Accident Forgiveness in public cloud billing, emphasizing the financial risks associated with cloud services and how Fly.
Thomas Ptacek
8 min read
Has Summary
--