The AI CLI split-pane workflow is the detail that stood out to me. I've been running Claude Code in Windows Terminal and the context-switching is genuinely annoying — you're constantly tabbing between the agent output and your shell. Having yaw auto-detect the tool and open a companion terminal in the same directory is the kind of small UX decision that compounds over a full workday. Tried it this morning. Works as described.
The connection manager is also underrated in the post. Encrypted credential storage + Tailscale auto-detection in one place covers a lot of ground for anyone managing a handful of SSH hosts and databases regularly.
Question for the dev: how are credentials encrypted at rest? Local keychain, or something custom?
Comments
The AI CLI split-pane workflow is the detail that stood out to me. I've been running Claude Code in Windows Terminal and the context-switching is genuinely annoying — you're constantly tabbing between the agent output and your shell. Having yaw auto-detect the tool and open a companion terminal in the same directory is the kind of small UX decision that compounds over a full workday. Tried it this morning. Works as described. The connection manager is also underrated in the post. Encrypted credential storage + Tailscale auto-detection in one place covers a lot of ground for anyone managing a handful of SSH hosts and databases regularly. Question for the dev: how are credentials encrypted at rest? Local keychain, or something custom?
yeah, the ai cli split-pane workflow really works for me. creds are encrypted via local keychain.