Manus is so popular that MetaGPT is open-sourced again. OpenManus-RL introduces reinforcement learning; multi-agent collaboration framework OWL

✨ 1: OpenManus-RL

OpenManus-RL is an open source project jointly developed by UIUC and MetaGPT, which aims to explore a new paradigm for tuning large language model agents based on reinforcement learning.

OpenManus-RL is an open source project jointly led by Ulab-UIUC and MetaGPT, which aims to use reinforcement learning (RL) to improve the reasoning and decision-making capabilities of large language models (LLMs) as intelligent agents. Inspired by successful cases such as Deepseek-R1 and QwQ-32B, the project explores a new paradigm for tuning LLM agents based on RL.

Address: https://github.com/OpenManus/OpenManus-RL/blob/main/Readme.md

✨ 2: mcp-server-chatsum

mcp-server-chatsum is an MCP server that helps users understand the content of conversations by querying and summarizing chat logs.

MCP-Server-Chatsum is a MCP (Model Context Protocol) server that is specifically designed to summarize your chat history. It can query chat messages with specified conditions and summarize these messages based on the query prompt. This server is designed to integrate with applications such as Claude Desktop to provide them with contextual information, thereby enhancing the intelligence of the conversation.

Address: https://github.com/chatmcp/mcp-server-chatsum

✨ 3: OWL

OWL is a multi-agent collaborative framework that enables general automation assistance for real-world tasks by optimizing workforce learning.

OWL (Optimized Workforce Learning) is a multi-agent collaboration framework built on the CAMEL-AI framework, which aims to push the boundaries of task automation. Its vision is to revolutionize the way AI agents collaborate to solve real-world tasks.

Address: https://github.com/camel-ai/owl

✨ 4: Nanobrowser

Nanobrowser is an open-source AI web automation tool that runs in your browser, is free, and privacy-focused.

Nanobrowser is an open source AI web automation tool that runs as a Chrome browser extension and is designed to provide powerful web automation capabilities, completely free of charge. Compared to paid services such as OpenAI Operator, the advantage of Nanobrowser is that it allows users to fully control their own data and API keys, and all operations are performed in the local browser to ensure privacy and security. It supports multiple large language models (LLMs) and uses a multi-agent system to work together to complete complex network tasks.

Nanobrowser's main features include:

• Completely free:
   There are no subscription fees, users only pay for the API keys they use.
• Privacy protection:
   Everything runs locally in your browser, and data and credentials are not uploaded to the cloud.
• Flexible LLM options:
   Supports connecting to different LLM providers and assigning different models to different agents.
• Fully open source:
   All codes are transparent and users can understand the specific process of web page automation.
• Multi-agent systems:
   Different agents work together, such as Planner (strategy formulation), Navigator (webpage navigation) and Validator (verification of task completion).

Address: https://github.com/nanobrowser/nanobrowser

✨ 5: Local-NotebookLM

Local-NotebookLM is a local AI tool that converts PDF to podcasts, supports multiple LLM and TTS models, and provides an API interface.

Local-NotebookLM is a local AI-powered tool that can convert PDF documents into engaging podcasts. It uses local LLM (Large Language Model) and TTS (Text-to-Speech) models, so it can run locally without relying entirely on cloud services.

Key features:

• PDF Processing:
   Extract text from PDF, clean it and format it.
• Podcast Generation:
   Generate customizable podcasts in different styles (casual, formal, technical, academic) and lengths (short, medium, long, very long), supporting multiple formats (podcast, article, summary, interview).
• LLM supports:
   Supports multiple LLM providers, including OpenAI, Groq, LMStudio, Ollama, Azure, etc. This means you can choose the model you like, or use a locally deployed model to reduce costs and protect privacy.
• TTS conversion:
   Supports text-to-speech, and you can choose different voices.
• Highly Configurable:
   The entire process is highly configurable, including models, parameters, etc.
• API:
   Provides a programmatic API for easy integration into other projects.
• FastAPI service:
   Provides a FastAPI server that can be accessed using a web interface.

Address: https://github.com/Goekdeniz-Guelmez/Local-NotebookLM