3 ms·

Hey Hacker News, I’m excited to share LangManus, an open-source project that’s all about community-driven AI automation. Built on the shoulders of amazing open-

by anxs 2y ago

Hey Hacker News, I’m excited to share LangManus, an open-source project that’s all about community-driven AI automation. Built on the shoulders of amazing open-source work, LangManus combines language models with specialized tools to tackle tasks like web search, crawling, and Python code execution—while giving back to the community that inspires it. At its core, LangManus uses a hierarchical multi-agent system, where a supervisor coordinates a team of specialized agents to handle complex workflows. It integrates with LLMs like Qwen and powers up with tools like Tavily for search and Jina for crawling. Plus, it comes with Docker support and a web UI to make interaction a breeze. Want to see it in action? Check out the demo video: https://youtu.be/sZCHqrQBUGk https://youtu.be/sZCHqrQBUGk. In it, LangManus calculates the influence index of DeepSeek R1 on HuggingFace using a fully automated plan—think web searches, data retrieval, and Python code crunching, all seamlessly tied together. Here’s what makes LangManus stand out: Multi-Agent Magic: A supervisor delegates tasks to agents like Researcher, Coder, and Browser for efficient execution. Tool Integration: Web search with Tavily, neural search with Jina, and Python REPL for coding on the fly. Ease of Use: Docker-ready and a web UI for quick setup and control. If you’re into AI automation, multi-agent systems, or just love contributing to open-source projects, swing by the GitHub repo: https://github.com/langmanus/langmanus https://github.com/langmanus/langmanus. Dive into the code, play with the demo, or drop some feedback—we’d love to hear from you. Join the community and let’s build something awesome together!

shrisukhani 2y ago
Super cool. Congrats on the launch! Let me know if we can help with something at Hyperbrowser.
cafed00d 2y ago
How do you think about core business logic (or at least, _significant_ business logic) being embedded within prompts like here: https://github.com/langmanus/langmanus/blob/main/src/prompts/researcher.md?plain=1 https://github.com/langmanus/langmanus/blob/main/src/prompts... Do you think or worry about not-being able to test these things? (Or is that just me :)) Details: I ack/understand this comes from a dependency (ReAct agents); not directly langmanus. But, still, curious what the community/hn-tech thinks of testability, veracity, potentially conflicting or overlapping instructions across agents, etc, wrt “prompts” as sources of logic. Ack its a general practice with LLMs.
- cship2 2y ago
  Thought the logic was going to go with cpu and inference related tasks will use GPU.

by anxs 2y ago

shrisukhani 2y ago

Super cool. Congrats on the launch! Let me know if we can help with something at Hyperbrowser.

cafed00d 2y ago

How do you think about core business logic (or at least, _significant_ business logic) being embedded within prompts like here: https://github.com/langmanus/langmanus/blob/main/src/prompts/researcher.md?plain=1 https://github.com/langmanus/langmanus/blob/main/src/prompts... Do you think or worry about not-being able to test these things? (Or is that just me :)) Details: I ack/understand this comes from a dependency (ReAct agents); not directly langmanus. But, still, curious what the community/hn-tech thinks of testability, veracity, potentially conflicting or overlapping instructions across agents, etc, wrt “prompts” as sources of logic. Ack its a general practice with LLMs.

cship2 2y ago
Thought the logic was going to go with cpu and inference related tasks will use GPU.