Cindy is an open-source AI agent developed in TypeScript, designed to operate locally using real files and applications. The project, hosted in a pnpm monorepo, includes Electron desktop and Expo/React Native mobile clients alongside shared packages for authentication, device linking, and model orchestration. It supports multiple harnesses—including Claude Code and Codex—with the ability to switch between models mid-task while maintaining continuous workspace, memory, and tool integration. Technical requirements include Node.js 22.x and pnpm 10.x. The system enables parallel task execution across harness-model combinations and integrates browser, computer, and phone automation. TapDB analytics are included by default but can be disabled via settings, tracking aggregate usage metrics without collecting chat or file content. Contributions require adherence to Apache-2.0 licensing, Developer Certificate of Origin sign-offs, and workflows outlined in CONTRIBUTING.en.md. The client connects to Cindy’s cloud services by default, with optional "Skip Sign-In" mode for local-only functionality. Installation guides for Linux (Ubuntu, Arch, Omarchy) detail setup, updates, and keyring configuration. The project’s architecture supports extensibility through plugins, MCP integrations, and shared skills, with self-hosted backend services maintained in a separate repository. TapDB’s data collection includes device identifiers and OS versions, with explicit consent mechanisms for analytics.