DeepChat

A local-first AI agent desktop client that integrates multiple LLM providers with MCP and skill support.

DeepChat screenshot 1

DeepChat is an open-source desktop app that functions as a local-first AI agent desktop client. It provides a unified interface for interacting with cloud APIs and locally deployed models, allowing users to manage multiple AI assistants within a single application. The software is available for Windows, macOS, and Linux, serving as a central hub for various large language model providers.

The application focuses on agent capabilities and session persistence. It allows users to run long-lived sessions with project folders, specific permission modes, and resumable context. It integrates with various messaging platforms, enabling users to control their desktop AI sessions remotely through services like Telegram, Discord, Feishu, and WeChat iLink. This allows for a hybrid workflow where AI agents can be managed from a mobile device while running on a local machine.

Key features

  • Support for cloud LLMs including OpenAI, Gemini, Anthropic, and DeepSeek
  • Integrated Ollama management for local model deployment and execution
  • Model Context Protocol support for Resources, Prompts, and Tools
  • Session Tape recording for recoverable and traceable work history
  • Installable Skills for task-specific instructions and reference files
  • Agent Client Protocol integration to run coding and task agents
  • Multi-window and multi-tab architecture for parallel sessions
  • Built-in search extensions for Google, Bing, and Brave Search
  • Artifacts rendering and Mermaid diagram support

DeepChat is built with Electron and follows the Tape.systems philosophy to ensure that tool calls and request sequences remain inspectable. It is designed for users who require high levels of control over their AI workflows, including developers who use MCP services or those who need to import skills. The architecture separates data interaction from UI behavior to maintain performance and avoid the limitations of simple web wrappers. It includes a built-in Node.js runtime to ensure that npx and node-style services work without additional configuration.

This tool is a privacy-focused, business-friendly client for users who want to consolidate multiple AI providers into one local interface while maintaining full ownership of their session data.

Last Modified
Software TypeDesktop App
Platform
Last Activity14 days ago
Repository Age1 year
LicenseApache-2.0
Open Source Alternative to
Open Source Software.io

Join our newsletter to get shiny new open source software delivered to your inbox. Unsubscribe anytime.