About CodePilot
CodePilot is a desktop AI workspace that brings Claude Code, Codex and its own built in CodePilot agent into a single application for project work and everyday tasks. Each new conversation is assigned to one agent, runs with its own independent context, and can open a project workspace where the user inspects files and reviews changes live as the agent works.
It is aimed at developers who already use coding agents and want one place to run several sessions at once. Permission control sets how an agent asks for approval, sessions persist across restarts, and a personal assistant workspace keeps preferences and memories. Agents can be extended with Skills, MCP servers and command line tools, and a guided setup connects services. Model connections are flexible: a supported subscription sign in, API keys for Anthropic, OpenRouter, DeepSeek, GLM or Kimi, any Anthropic compatible or OpenAI compatible endpoint, or local models through Ollama. The built in agent works without an extra CLI, while the Claude Code and Codex options need their own CLIs installed.
CodePilot does not publish a price list. The source is released on GitHub under the Business Source License 1.1, and personal, educational, nonprofit and evaluation use is free. A separate commercial license is required for paid products or services offered to third parties and for internal use by organizations with more than 100 employees. Model usage is billed separately by whichever AI provider the user connects.
One limitation to note: a conversation stays with the agent chosen at the start, so switching between Claude Code, Codex and CodePilot means opening a new conversation, although compatible models and providers can be changed within the same agent.