Containers
A container in CropsCode is an isolated development environment where an AI agent works on a task. Each container gets its own filesystem, network, and runtime — like giving every ticket its own machine.
Lifecycle
Section titled “Lifecycle”| Phase | Farm Term | What Happens |
|---|---|---|
| Launch | Planting | Container created from image, repo cloned, agent starts |
| Running | Growing | AI agent works on the task, you can watch via terminal |
| Complete | Harvest | Work is done — PR drafted, branch pushed, tests passing |
| Stopped | Cleared | Container removed, resources freed |
What’s Inside a Container
Section titled “What’s Inside a Container”Every container includes:
- An AI coding agent (Claude Code, GitHub Copilot, or Codex)
- Git, GitHub CLI, tmux, neovim
- The cloned repository with ticket context
- A generated
CLAUDE.mdwith task instructions - Network access for package installs and API calls
Runtime Images
Section titled “Runtime Images”CropsCode uses Podman container images to define what’s inside each container. The default image (claude-code-terminal) includes everything needed for general development.
You can build custom images for specialized environments. See Custom Containerfile.
Container Limits by Plan
Section titled “Container Limits by Plan”| Plan | Concurrent Containers | History |
|---|---|---|
| Starter | 5 | 14 days |
| Pro | 15 | 90 days |
| Enterprise | Unlimited | 365 days |
Interacting with Containers
Section titled “Interacting with Containers”From the orchestrator dashboard:
- Live terminal — web-based terminal (xterm.js) connected directly to the container
- Resource monitoring — CPU, memory, disk usage in real time
- VS Code — attach VS Code to a running container for full IDE access
See VS Code Extension for IDE setup.