This is an early release preview. You may encounter bugs.
DeepChat logo
Unclaimed

App interface connectors

DeepChat

Local-first AI agent desktop client with MCP, Skills, ACP agents, and remote control from messaging apps

01 / Summary

About DeepChat.

DeepChat is a local-first desktop client for chatting with LLMs and running agents on Windows, macOS, and Linux. It brings cloud models, local Ollama models, MCP tools, Skills, external agent runtimes, and long-running sessions into one Electron app.

Sessions follow the Tape.systems approach: each session records a structured Tape of context, tool calls, requests, and results so work can be recovered, resumed, and inspected. Trace previews show request sequences, provider and model metadata, which entries were included, and token budgets.

DeepChat supports the Agent Client Protocol, so ACP-compatible coding and task agents appear as entries in the model selector next to ordinary models, with a workspace view for their plans, tool calls, and terminal output. Skills follow the Agent Skills specification and import or export with Claude Code, Codex, Cursor, Windsurf, GitHub Copilot, OpenCode, Goose, Kilo Code, and other compatible tools.

Model providers include DeepSeek, OpenAI, Moonshot (Kimi), Grok, Gemini, Anthropic, Zhipu, MiniMax, Azure OpenAI, OpenRouter, LM Studio, and many relay services, plus any provider that speaks the OpenAI, Gemini, or Anthropic API format. Sessions can also be driven remotely from messaging apps.

  • ACP agents: built-in or custom ACP-compatible agents run as selectable models with a dedicated workspace view
  • MCP support: Resources, Prompts, and Tools over StreamableHTTP, SSE, and Stdio, with in-memory services for code execution, web retrieval, and file operations
  • Skills: install from folders, ZIP files, or URLs, enable per conversation, and import or export with other agent tools
  • Built-in Skills: code review, document collaboration, DOCX, PDF, PPTX, XLSX, frontend design, MCP building, and Skill creation
  • Tape and Trace: recoverable session history and request traces with token budgets
  • Remote control: drive sessions from Telegram, Feishu/Lark, QQBot, Discord, and WeChat iLink, including model switching and permission prompts
  • Local models: Ollama model downloads, deployments, and runs managed from the app
  • Search: BoSearch and Brave Search APIs plus browser-simulated search on Google, Bing, Baidu, and Sogou
  • Chat interface: multi-window and multi-tab sessions, forkable conversations, Artifacts, Mermaid diagrams, and image rendering
  • DeepLinks: start conversations or install MCP services from a link
  • Privacy controls: local data storage, network proxy support, and screen projection hiding

02 / Discussion CREDIBILITY-GATED

Discussion

Reading is open to everyone. Posting and voting need a verified identity or a GitHub grade of B or higher.

  • No discussions yet.

04 / Build

Build with DeepChat.

Browse the catalog for frameworks, tools, and harnesses, each scored on real GitHub credibility.

Get DeepChat →

Browse the catalog