I went quiet here after 0.22. We're on 0.25.5 now, and most of what landed in between is the unglamorous stuff: speed, reliability, and getting out of your way. Here's what actually changed if you've been away.
Two releases since split panes landed. Here's what shipped.
v0.27 Pins & Notes
You know the "found this, not ready to fix it yet" pile? Now you can pin any terminal output to a per-project rail and work it as a checklist. Title + body per pin, drag to reorder, done items collapse. Works inside full-screen TUIs like Claude Code, not just plain shells. A notes scratchpad sits under it and autosaves per project.
Ten days ago we launched on Product Hunt. Since then v0.22 dropped. Probably the biggest one yet. Quick rundown:
Agent instructions library
Pick a role (Software Architect, Code Reviewer, Debugger, Refactoring Specialist, Test Author, plus 8 more) and the system prompt loads natively. --append-system-prompt for Claude, --system-instruction for Gemini. Stop pasting the same prompt into every new session.
Running 5 AI agents across 3 projects? Tabs everywhere, no idea what each one is doing.
HiveTerm is one terminal workspace where Claude Code, Codex, Gemini and your dev stack run side by side. Config-driven (hive.yml), per-agent recap in the sidebar, voice input, and an MCP server so agents can spawn and talk to each other.
Free to start — up to 3 projects on the free tier. macOS, Windows, Linux.
HiveTerm is a desktop terminal that runs your AI agents (Claude Code, Codex, Gemini, Cline, etc.) alongside servers, builds, and scripts, all orchestrated by a hive.yml file. What sets it apart is Queen, an integrated MCP server that allows agents to create sub-agents, read output from other processes, and send notifications to one another. Add to that per-process resource monitoring, Git integration with a diff viewer, automatic stack detection, and context-separated workspaces.
