Klaussy Desktop

Klaussy Desktop

**The agent-first IDE that reviews its own code before you commit.**

[](https://github.com/steph-dove/klaussy-desktop/stargazers) [](https://github.com/steph-dove/klaussy-desktop/releases)

**The desktop is the face. The [`klaussy-agents`](https://github.com/steph-dove/klaussy-agents) engine is the spine.** While other orchestrators give you a worktree grid, Klaussy brings the conventions, custom skills, and active git-level guardrails that actually keep agents aligned with your codebase.

Designed by an ex-GitHub, ex-Twitch, and ex-Microsoft engineer, Klaussy Desktop runs multi-agent, multi-repository AI coding sessions in a unified, tabbed light-themed workspace. It spawns isolated Git worktrees per task, enabling you to run multiple agents in parallel without juggling branches, staging mess, or local workspace clutter.

**📣 Out of stealth.** Klaussy has been in heads-down development for six months — quietly battle-tested by a private group of developers shipping real production work through it every day. After months of dogfooding, refinement, and distributing builds to testers, it's finally open to everyone.

---

🚀 Key Highlights

• ⚡ Parallel Agents, Isolated Worktrees One task = one git worktree + one independent agent. Run Claude Code, Codex, Gemini, Antigravity, Copilot, Cursor, Cline, opencode, and Kimi Code side by side — or drive a tab with a local model through Ollama (via Aider), or drop to a plain shell. The visual sidebar tracks staged/unstaged counts, status, and branch alignment per task.
• 🧪 Sandboxed Agents (Nemesis8) Optionally run an agent inside an isolated [Nemesis8](https://github.com/DeepBlue