LangBot
langbot-app/LangBotReviewed by AI
Checked by RepoPicks Radar on
What is LangBot?
LangBot is a server you run yourself that turns an AI model into a chat bot on messaging apps: Discord, Telegram, Slack, LINE, QQ, WeChat, WeCom, Lark, DingTalk, KOOK and Matrix. You choose a model (OpenAI, Anthropic, DeepSeek, Gemini, or a local one through Ollama), can feed it your documents as a knowledge base, and manage it all in a browser panel instead of editing files. It starts with one command through uv, or with Docker Compose. You still need a bot account on each chat app and, for most models, a paid key.
How easy is it to install LangBot?
C · install 1 or 2 tools · Windows
- Windows C
- macOS C
- Linux C
Windows, Mac and Linux take the same route and get the same grade: install the uv tool, run "uvx langbot", then open http://localhost:5300 and finish in the browser panel, where you paste the model key and connect each chat app. Docker Compose is the other documented route and is heavier. There is no installer to double-click, and no release carries a binary.
LangBot on GitHub
- Stars
- 18,883on Nov 10, 2026
- Downloads
- 31,879787 this week
- Growth
- +154stars in 6 days
Stars over time
Facts
- Runs on
- Windows, macOS, Linux
- You need to install
- Python
- To build from source
- Python, Docker
- Disk space
- 100 MB–300 MB · estimate
- Accounts or keys
- A model provider such as OpenAI, Anthropic or DeepSeek (a local model through Ollama or LM Studio replaces it), A bot account on each chat platform you connect (Discord, Telegram, Slack, WeCom, Lark, DingTalk and so on)
- Cost
- Free, API costs extra · The program is free under Apache-2.0. The models usually cost per use (OpenAI, Anthropic, DeepSeek and the listed gateways); Ollama or LM Studio on your own machine avoids that. The README also points to a version hosted by the maintainers: "LangBot Cloud (Recommended) ... Zero deployment, ready to use" at space.langbot.app/cloud.
- License
- Apache-2.0 (Permissive)
- Last change on GitHub
- Nov 10, 2026
LangBot risk signals (4)
- Attentionruns code and tools the model chooses
The Dockerfile builds Google's nsjail so "the Box runtime isolate code without a host Docker socket" and installs Node "so the sandbox (nsjail/Docker box) can run npx-based stdio MCP servers". So the bot can execute code and launch outside tools during a conversation; the Docker CLI is also shipped for the optional box service that "can drive the mounted host Docker socket", which hands the container control of Docker on your machine if you turn that on.
AI review
- Attentionplugins from a marketplace
The README advertises "Hundreds of plugins" and a Plugin Market at space.langbot.app, and plugins/README.md is the "Official plugin source index" with the note that "Plugins are installed separately and are not bundled in Core releases". Plugin code you install comes from third parties and runs inside your bot.
AI review
- Attentionconnects personal wechat and qq accounts
The supported-platform table lists "WeChat | Personal & Official Account" and "QQ | Personal & Official API (Channel, DM, Group)", and the dependencies include "gewechat-client" and "nakuru-project-idk". So the bot can be attached to a personal account and not only to an official bot account; each chat service sets its own rules for that.
AI review
- Infomanagement api also open to assistants
The README says LangBot "exposes a built-in Model Context Protocol endpoint at /mcp, mirroring the HTTP API so an agent can manage bots, pipelines, plugins, and models programmatically" and that you "Authenticate with the same API key ... no login flow required". Anyone who gets that key can manage the whole installation.
AI review
Evidence is shown in its original language.
What Radar could not check about LangBot
- Whether the connector for each chat app still works, since those platforms change their rules and APIs often
- What the model calls would actually cost for your message volume
- What LangBot Cloud, the hosted version the README recommends, includes or charges
- Whether a personal WeChat or QQ account stays usable while a bot drives it
- Real disk and memory use once a knowledge base with its vector database is filled
- What any plugin or MCP server you install from the market does inside your bot