Bilibili-first bot framework with a ChatGPT backbone
Because manually replying to every Bilibili @-mention is no way to live.

What it does
Langup is a Python toolkit that glues OpenAI’s API (via LangChain) to Bilibili’s platform APIs, letting you deploy LLM-driven personas across livestreams, video comments, direct messages, and voice chat. You define a system prompt—say, a sarcastic livestreamer or a helpful summary bot—and the library handles listening, filtering, and replying within Bilibili’s ecosystem. It also supports terminal and speech-based interaction if you want to test your persona offline.
The interesting bit Most LLM bot frameworks treat Twitter or Discord as the default stage; Langup treats Bilibili as the first-class citizen, complete with browser-cookie credential sniffing, content filtering for live barrage, and “at-signal” parsing in video threads. The VTuber live-chat mode is the standout: it ingests bullet comments, runs them through a persona, and spits replies back into the stream with concurrency controls.
Key highlights
- First-class Bilibili integration: livestream chat, video @-replies, and DM bots out of the box.
- Persona-driven: behavior is shaped entirely by a
systemprompt string, from “helpful assistant” to “angry contrarian.” - Automatic credential extraction from recently used browser sessions (Edge, etc.), avoiding manual cookie copy-paste.
- Voice input support via speech recognition, plus a plain terminal mode for local testing.
- Built-in content filtering and configurable ban-word lists for livestream safety.
Caveats
- Several listener modules—including WeChat, QQ, and Bilibili private messages—are marked as unfinished in the todo list.
- The README itself warns that the code may throw exceptions, and the architecture diagram notes some modules are still pending implementation.
- Platform scope is narrow: outside of Bilibili and generic console/voice modes, there is little here for other networks.
Verdict Chinese streamers and Bilibili UP hosts who want to experiment with LLM automation will find this a convenient starting point. If you have no presence on Bilibili, there is almost nothing here for you.
Frequently asked
- What is jiran214/langup-ai?
- Because manually replying to every Bilibili @-mention is no way to live.
- Is langup-ai open source?
- Yes — jiran214/langup-ai is open source, released under the MIT license.
- What language is langup-ai written in?
- jiran214/langup-ai is primarily written in Python.
- How popular is langup-ai?
- jiran214/langup-ai has 728 stars on GitHub.
- Where can I find langup-ai?
- jiran214/langup-ai is on GitHub at https://github.com/jiran214/langup-ai.