browser-use/browser-use
Browser-use is a project that enables AI agents to control and automate web browsers, allowing users to perform complex tasks through natural language commands and AI-driven browser interactions.
Awesome AI Agents › Browser Automation Tools
McGill-NLP/webllama
WebLlama is a project focused on building advanced AI agents powered by Meta's Llama 3 model that can browse the web by following user instructions and engaging in dialogue. The project aims to create human-centric web browsing assistants that enhance user experience rather than replace users. The core of WebLlama is a fine-tuned Llama-3-8B-Web model trained on the WebLINX 1.0 dataset, which includes over 24,000 curated instances of web interactions such as clicks, text inputs, submissions, and dialogue acts. This model surpasses GPT-4V by 18% in zero-shot performance on the WebLINX 1.0 benchmark, demonstrating superior ability in selecting useful links, clicking relevant elements, and generating aligned responses. WebLlama provides training scripts, optimized configurations, and instructions for training cutting-edge Llama agents on web navigation tasks. It also offers benchmarks for evaluating these agents on real-world web browsing scenarios, including multi-turn dialogue-based navigation and automatic web navigation. The project emphasizes continuous data curation and expansion, incorporating datasets like Mind2Web to improve agent generalization across millions of websites. Deployment is facilitated through integration with popular platforms such as Playwright, Selenium, and BrowserGym, making it easier to use these agents in practical applications. WebLlama also supports systematic evaluation with tools to visualize results and compare performance across different models and benchmarks. The project is open-source and provides easy access to models and datasets via Hugging Face, along with detailed documentation and example usage code. Overall, WebLlama represents a significant advancement in AI-driven web navigation, combining state-of-the-art language models with real-world interaction data to create powerful, user-friendly browsing assistants.
https://github.com/McGill-NLP/webllama
Also in Browser Automation Tools
unclecode/crawl4ai
Open-source Python web crawler and scraper that converts web pages into clean, LLM-ready Markdown for RAG pipelines and AI agents, with an API and an MCP server.
CloakHQ/CloakBrowser
CloakBrowser provides a stealthy Chromium binary with C++ source-level fingerprint patches to pass bot detection, serving as a drop-in Playwright/Puppeteer replacement.
feder-cr/dots
dots is an open-source CLI/web agent pairing any OpenRouter model with a patched Firefox engine for undetectable, human-like browsing, with per-seed identities and persistent login profiles.
jackwener/OpenCLI
OpenCLI converts any website or local tool into a command-line interface, enabling both human interaction and AI agent automation through logged-in browser sessions.
browserbase/stagehand
Stagehand is a production-ready AI web browsing framework that combines code and natural language to enable reliable and extensible browser automation.
lexmount/moli
Rust headless browser built for AI agents, with on-demand layout and rendering plus a CLI and CDP/WebDriver Classic/BiDi servers for lightweight page fetching and browser-use workloads.
citrolabs/ego-lite
ego lite is a specialized browser designed for AI agents to perform web automation in parallel with human browsing, offering isolated workspaces and enhanced page snapshots.