Features
Multi-project sessions with per-session AI model, mode, permissions, cwd, and history
Native Claude Agent mode with thinking blocks, tool calls, and diff rendering
Terminal mode supporting Codex, Gemini, shell, and custom scripts
First-class git worktree support for isolated branch sessions
Session switcher with ⌘1..⌘9 keyboard shortcuts
Per-session permission levels: manual, assisted, autonomous
Embedded terminal with slash commands (e.g., /agents, /remote-control, /compact)
Image attachment with thumbnail preview above composer
Prompt composer with stackable roles, style levels, and templates
Stable expand button on long code blocks with no drift
Stop button reliably cancels agent streams
Design system v2 with voice colors, composer refresh, status bar overhaul
Theme system v2 with 8 themes, including mono-tone variants
Branch picker single-click selection
Recover agent sessions from prompt-too-long errors
Multi-agent orchestration with handoffs
Sandbox Agents for containerized long-running tasks
Agent-as-tool delegation (v0.15.0+)
Realtime Agents with gpt-realtime-2 voice support (v0.17.6+)
Human-in-the-loop mechanisms
Automatic session history management
Built-in tracing for debugging
Provider-agnostic LLM support (100+ models via LiteLLM)
Redis session support (optional, v0.17.6+)
Instructions, tools, and guardrails configuration
Jupyter notebook compatibility
Supports OpenAI Responses and Chat Completions APIs
