Cindy

Cindy is a local AI agent for people who want to delegate work across code projects, files and connected apps. It can carry a task across different coding agents and models while retaining its context.

• Agent switching: Switch between Claude Code, Codex and Pi during a task while keeping the same workspace, memory, skills and tools.
• Parallel work: Assign planning, execution and review to separate agents using different models, with shared context and acceptance criteria.
• Projects and apps: Work with local files, terminals and signed-in browser sessions, connect services such as GitHub and Jira, or reach internal systems through MCP.
• Review and recovery: Inspect edits line by line, approve sensitive actions and return to checkpoints; code and file changes begin in isolated workspaces.
• Remote and recurring tasks: Follow sessions from mobile, delegate through Slack or Feishu, and schedule jobs that run on your computer and report back to chat.

The free, open-source client uses Apache-2.0 and runs on macOS, Windows and Linux. Companion apps for iOS, iPadOS and Android let you view and control desktop sessions. Model connections include Ollama, vLLM and OpenAI-compatible endpoints, your own provider keys, existing coding subscriptions, or Cindy's optional paid managed service.

Files, code and chat history stay on your computer by default, but inference requests go to the model endpoint you choose. Local agents don't require a Cindy account; server-backed features require sign-in. Official builds include usage analytics that exclude chat and file contents.