Problem solved

Developers using multiple AI CLI tools like Claude Code, Codex, and Gemini often struggle with fragmentation, having to switch between terminals and manually manage separate sessions. This leads to inefficiency, loss of context, and an inability to leverage multiple agents together effectively. Dorothy solves this by providing a unified desktop environment where these agents can be orchestrated concurrently, with built-in task management and automation, reducing context-switching and enabling complex, multi-step coding workflows that would otherwise require manual coordination.

Use cases

• Running Claude Code, Codex, and Gemini simultaneously to compare code completions and choose the best output.
• Creating an automated pipeline where one agent writes unit tests and another agent reviews and refactors them.
• Using the Kanban board to track feature development tasks, each assigned to a specific AI agent for implementation.
• Remotely instructing an agent via a Telegram bot to fix a bug while away from the development machine.
• Integrating MCP servers to fetch real-time social media trends and have an agent generate relevant code snippets.
• Managing per-agent skill configurations for different coding environments like Claude Code and OpenCode.

Audience

• Developers
• Data Scientists
• Hobbyists

Pros

• Unifies multiple AI CLI agents in a single desktop interface, reducing context-switching.
• Visual Kanban task management streamlines AI-assisted development workflows.
• Extensible through MCP servers, enabling integration with external tools and services.
• Supports automations for complex multi-agent coding pipelines.
• Open-source and free, with community-driven development.

Cons

• Limited to supported CLI agents (Claude Code, Codex, Gemini); other agents require custom integration.
• Requires separate installation and setup of each underlying agent tool.
• Desktop-only application, lacking a web-based interface for remote access without MCP workarounds.