ИИ-агенты
go-micro
A framework for building agents and services
★ 23,1 тыс. звёздApache-2.0
Смотреть руководство →LaVague — проект с открытым исходным кодом из категории «ИИ-агенты» от lavague-ai (lavague-ai/LaVague на GitHub). У него 6 395 звёзд на GitHub и 575 форков, он распространяется по лицензии Apache-2.0 и последний раз обновлялся на GitHub 21 января 2025.
A Large Action Model framework for building AI web agents that browse and act for you.
Define the objective your agent should accomplish on a website, in one clear sentence.
Install LaVague from PyPI and pick a driver — Selenium, Playwright or the Chrome extension.
Connect the multimodal LLM the World Model will use to read pages and plan actions.
Launch the Gradio demo or write a short script, give the agent an objective, and watch it act on the page.
A World Model takes your objective plus the current web page state and outputs the set of instructions needed to progress.
Instructions are compiled into action code — Selenium or Playwright — and executed against the live page.
Supports a Selenium webdriver, a Playwright webdriver and a Chrome extension driver.
LaVague QA provides tooling aimed at QA engineers for automated web testing.
A ready-made Gradio agent demo lets you try web agents interactively without writing integration code first.
Data collection is documented in the README, including how to turn telemetry off entirely.
Clone LaVague to explore the framework and examples.
git clone https://github.com/lavague-ai/LaVague.git
cd LaVagueInstall LaVague from PyPI.
pip install lavagueStart the Gradio agent demo and give the agent a web objective (see the README's Getting Started for the exact demo command).
# Launch Gradio Agent Demo — see README for the exact commandLaVague is maintained by lavague-ai, with documentation at docs.lavague.ai.
The framework is Apache-2.0 licensed and free to use; you pay for the LLM calls your agents make.
Yes, in practice: the World Model uses an LLM (the README's examples use GPT-4-class multimodal models), so you need a key for your chosen provider.
Plain Selenium requires hand-written scripts; LaVague adds an LLM World Model that plans actions from a natural-language objective, then compiles them to Selenium or Playwright.
pip install lavague, configure your LLM, and run the Gradio demo with a simple objective.
A framework for building agents and services
Vercel's open framework for building agents that run anywhere.
Chinese guide to building production-grade AI agents with LangChain and LangGraph.
Qwen-AgentWorld: a language world model simulating MCP, terminal, web, and OS environments for agents.
A dev-first framework for building and running autonomous AI agents.
A framework for building realtime voice AI agents 🤖🎙️📹
Другие проекты из категории «ИИ-агенты» с наибольшим числом звёзд в каталоге LibreAI — реалистичный список альтернатив, если LaVague вам не подходит.
A personal AI that really does things — on any OS, with your data staying yours.
An agentic skills framework & software development methodology that works.
Agent harness optimization system with skills, memory and security for Claude Code, Codex and more
The agent that grows with you — a Python framework for adaptive AI agents.
DeepSeek's plugin-based agent harness where everything is a plugin.
Соседние проекты из категории «ИИ-агенты» в каталоге, ближайшие по числу звёзд к LaVague — продолжайте изучать полку по обе стороны от него.
Policy, identity, and sandboxing guardrails for AI agents.
A local-first AI agent workspace for coding, writing, design, research and automation.
An MCP server and CLI that lets AI agents build, test and run iOS and macOS projects.
A control plane with a durable state kernel that keeps long-horizon agents moving across sessions.
Open-source framework for building agentic apps in JavaScript, Go, Dart, and Python, built and used in production by Google
Open-source LLM observability: monitor, evaluate, and experiment.
Бесплатно и открыто — забирайте исходный код, ставьте звезду репозиторию и начинайте создавать.