ИИ-агенты
WorkArena
WorkArena: ServiceNow's benchmark for how well web agents handle everyday knowledge-work tasks.
★ 275 звёздNOASSERTION
Смотреть руководство →ClawBench — проект с открытым исходным кодом из категории «ИИ-агенты» от TIGER-AI-Lab (TIGER-AI-Lab/ClawBench на GitHub). У него 975 звёзд на GitHub и 62 форка, он распространяется по лицензии Apache-2.0 и последний раз обновлялся на GitHub 5 октября 2026.
An open benchmark for browser AI agents on everyday tasks.
Start with one concrete task rather than a grand architecture.
Clone the repo and check the examples folder for a quickstart.
Install packages, set environment variables, and point it at your model.
Execute the demo workflow, then adapt it to your own data.
Daily chores, not toy problems.
Compare agents apples-to-apples.
Purpose-built for web agents.
Inspect and extend the tasks.
From TIGER-AI-Lab.
Free to use and adapt.
Download the source and move into the project folder.
git clone https://github.com/TIGER-AI-Lab/ClawBench.git
cd ClawBenchCreate a virtual environment and install the requirements.
python -m venv .venv && source .venv/bin/activate
pip install -r requirements.txtTreat these as starting points - the README documents the exact commands.
python examples/quickstart.pyTIGER-AI-Lab/ClawBench is its home on GitHub, with 62 forks so far. The contributors tab and recent commits show how actively it is maintained.
Yes - it is Apache-2.0 licensed, so it is free for commercial use with attribution.
Running the evaluation is free; the model under test may use your own paid API keys.
ClawBench is one specific benchmark for browser agents; Harbor is a framework for running eval programs. Run ClawBench inside Harbor-style workflows for the best of both.
Follow the README quickstart, then browse the examples directory for patterns that match what you want to build.
WorkArena: ServiceNow's benchmark for how well web agents handle everyday knowledge-work tasks.
OSWorld — benchmark multimodal agents on open-ended tasks in real computer environments.
Benchmark computer-use agents on long, real-world desktop tasks.
A benchmark testing whether agents can solve long-horizon coding tasks.
The open-source library that lets AI agents control a real browser.
A fast browser-automation CLI built for AI agents.
Другие проекты из категории «ИИ-агенты» с наибольшим числом звёзд в каталоге LibreAI — реалистичный список альтернатив, если ClawBench 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 ClawBench — продолжайте изучать полку по обе стороны от него.
An open-source speech language model from Hume.
An enterprise MCP gateway: governed, auditable tool access for agents.
Qwen-AgentWorld: a language world model simulating MCP, terminal, web, and OS environments for agents.
A memory and retrieval server built for LLMs.
A Python framework for building MCP servers and agent tools.
A self-improving QA agent that writes tests in plain English.
Бесплатно и открыто — забирайте исходный код, ставьте звезду репозиторию и начинайте создавать.