Developer Tools

Show HN: Coasts – Containerized Hosts for Agents

Open-source tool creates isolated development containers for AI agents without changing existing code

Deep Dive

Coasts has launched a new open-source CLI tool called Coasts (Containerized Hosts) that enables developers to run multiple isolated instances of full development environments on a single machine. The tool includes a local observability UI and works seamlessly with existing setups—requiring only a small Coastfile at the repository root. If you already use Docker Compose, Coasts can boot from your existing docker-compose.yml; if not, it works just as well without Docker. The system allows developers to build once and run N instances with whatever volume and networking topology their project needs, while maintaining complete isolation between environments.

Coasts is specifically designed for AI agent development workflows, remaining agnostic to AI providers and agent harnesses. The only host requirement is Git worktrees, allowing developers to switch between different AI tools without changing their workflow or requiring harness-specific environment setup. The platform is offline-first with no hosted service dependency, meaning there's no vendor lock-in risk—even if the Coasts team disappeared, local workflows would continue running. Currently macOS-first with Linux support, Coasts supports local HTTPS stacks using Caddy and provides certificate management for secure development environments. The tool enables developers to check out one coast at a time to bind canonical ports to their host, while using dynamic ports to peek into the progress of any worktree, making it ideal for testing multiple AI agent configurations simultaneously.

Key Points
  • Works with existing Docker Compose setups or without Docker—just add a Coastfile to repo root
  • Build once, run N isolated instances with custom volume/networking topologies for AI agent testing
  • Offline-first, provider-agnostic design eliminates vendor lock-in and works with any AI harness

Why It Matters

Enables scalable local testing of multiple AI agent configurations without cloud dependencies or vendor lock-in.