sbsh provides persistent terminal sessions with built-in discovery, environment profiles, and an API for automation.
*The problem:*
- Complex configuration required to access multiple Kubernetes clusters and Terraform workspaces, including setting visual prompts to identify production environments and avoid mistakes
- Manual setup of environment variables and credentials for each environment
- No easy, shareable way to reproduce those configurations across a team
- Lack of structured logs or visibility into existing sessions
- SSH sessions that die in the middle of debugging or long tasks
*How it works:*
sbsh separates the terminal session (your shell and environment) from the supervisor (the controller). Terminals continue running even if the supervisor stops or the network connection drops.
*Key features:*
- Terminal session discovery: sb get lists all sessions, sb attach mysession reconnects instantly
- Profiles: YAML-defined environments for Kubernetes contexts, Terraform workspaces, or Docker containers, identical in local dev and CI/CD
- Multi-attach: Several users can connect to the same live session
- API access: Control and automate sessions programmatically
- Structured logs: All input and output are recorded for replay or analysis
*Use cases:*
- DevOps: Persistent kubectl or Terraform sessions
- Developers: Long-running tests and builds over unstable connections; launching Python environments, npm projects, and more
- CI/CD: Identical profiles in local and pipeline environments
sbsh is a single Go binary (busybox-style) that can also be used as a login shell. It runs on Linux, macOS, and FreeBSD.
Repository: github.com/eminwux/sbsh
I have been using it for some time now, and it completely changed how I manage infrastructure. I would love to hear feedback and see how others might use it.