Inside a Pod
A Pod is one hosted Mac for one developer. It boots from a golden image in seconds, hosts a few agents at once, and is destroyed when the work is done.
The hardware
Pods run on dedicated Apple silicon Mac minis that Macswarm operates in a data center in the Netherlands (EU) or in the US. Each Mac mini runs two Pods. The machines are not shared with other providers’ workloads, and Enterprise customers can have a pool of Macs that is theirs alone.
What is installed
| Component | What you get |
|---|---|
| Xcode | The Latest Xcode and the previous major version, with command line tools |
| iOS runtimes | Current iOS plus the previous major, so apps can be checked on both |
| Simulators | Three to four devices pre-booted per Pod, leased to agents in turn |
| Agents | Claude Code and Codex, signed in through your connected environment |
| Tooling | XcodeBuildMCP for building, launching and driving the Simulator; git; Homebrew packages common in iOS projects |
| Build queue | Two concurrent xcodebuild runs, fair sharing between agents |
Other toolchains (Node, Python, Ruby for fastlane and CocoaPods) are present in the image. Tell us if your project needs something else in the base image.
How agents share a Pod
- Each session gets its own git worktree, so agents never step on each other’s changes.
- Builds enter a queue with two slots. An agent whose build is waiting keeps working on something else until a slot frees up.
- Simulators are leased per agent for the duration of a launch or a test run and returned to the pool.
- Two to three active agents per standard Pod are guaranteed. You can run more; builds and Simulator leases then queue.
Lifecycle
- A session starts and Macswarm hands out a Pod from the warm pool, cloned fresh from the golden image.
- Your repository is cloned, secrets are injected for the session, the agent starts working.
- More sessions join the same Pod while it is running.
- Ten minutes after the last session ends the Pod is destroyed. The next session gets a fresh clone.
Nothing persists on the Pod between sessions: not caches, not credentials, not build products. Derived data that matters lives in your repository or your CI cache.
Network
Pods make outbound connections only: to api.anthropic.com or api.openai.com, to your git host, and to package registries your build needs. There are no inbound ports and no SSH for customers. You watch agents through Claude, Codex, or the logs and screenshots in the dashboard.
Commits
Commits made by an agent are attributed to the developer who started the session, with the agent named in the trailer. That keeps blame and review history readable.
Next: Limits and queues.