SkillAgentSearch skills...

AWorld

Build, evaluate and train General Multi-Agent Assistance with ease

Install / Use

/learn @inclusionAI/AWorld
About this skill

Quality Score

0/100

Supported Platforms

Claude Code
Cursor

README

<div align="center">

AWorld: The Agent Harness for Your World

</div> <h4 align="center">

"The Next Frontier for AI is Your Expertise"

[![Twitter Follow][twitter-image]][twitter-url] [![WeChat QR Code][wechat-image]][wechat-url] [![Discord][discord-image]][discord-url] [![License: MIT][license-image]][license-url] [![DeepWiki][deepwiki-image]][deepwiki-url] [![Tutorial][tutorial-image]][tutorial-url]

<!-- [![arXiv][arxiv-image]][arxiv-url] --> <!-- [![Playground][playground-image]][playground-url] --> </h4> <h4 align="center">

中文版 | Automation | Manual | Evolution | Contributing |

</h4>
<p align="justify"> General AI often hits a "wall of context"—the nuanced data, workflows, and intuition that define <em>your</em> world. An agent's true power lies not in the model alone, but in its <b>Agent Harness</b>: the framework orchestrating its tools, memory, context, and execution.

This is the <b>AWorld Thesis</b>: A powerful harness is not enough. True AI scaling is unlocked only when experts like you embed the invaluable knowledge, effectively building the gate in that wall.

AWorld is the platform designed for this singular purpose. We provide a complete, battle-tested Harness as the recipe for you, the expert, to forge your knowledge into a fleet of autonomous agents. Together, we move beyond AI's generic promise to create robust, precise applications that master <em>your</em> specific domain.

</p>

From Expertise to Product

See what happens when expert knowledge is encoded into reusable Skills. The creations below are orchestrated by the AWorld Agent, demonstrating our core scaling law: as the community contributes more expertise, the entire ecosystem becomes more powerful.

This is what's possible today. Imagine what we'll build with your expertise.

<table> <colgroup> <col style="width:15%"> <col style="width:40%"> <col style="width:22%"> <col style="width:23%"> </colgroup> <thead> <tr> <th>Capability</th> <th>Expertise</th> <th>See it in Action</th> <th>Recipe</th> </tr> </thead> <tbody> <tr> <td>Create App</td> <td>• Auto-creation by base model<br>• Auto-evaluation by <a href="aworld-skills/app_evaluator/SKILL.md">UI Evaluation Skill</a></td> <td style="width:22%"><img src="readme_assets/aworld_cli_app_create.gif" alt="App create demo" width="270"></td> <td><a href="docs/Recipe/miniapp_build_recipe.md">View Recipe</a></td> </tr> <tr> <td>Create Video</td> <td>• Auto-creation by <a href="https://www.skillhub.club/skills/remotion-dev-remotion-remotion">Remotion Skill</a><br>• Human evaluation</td> <td style="width:22%"><img src="readme_assets/aworld_cli_intro_fast.gif" alt="Video create demo" width="270"></td> <td><a href="docs/Recipe/video_create_recipe.md">View Recipe</a></td> </tr> </tbody> </table>

Your Journey with AWorld-CLI

The journey from an idea to an evolved, autonomous agent begins at your fingertips.

Install and Activate

Install once, configure globally, and run anywhere.

Install AWorld-CLI

git clone https://github.com/inclusionAI/AWorld && cd AWorld

conda create -n aworld_env python=3.11 -y && conda activate aworld_env 

pip install -e . && cd aworld-cli && pip install -e .

Config & Launch

cd your working directory

aworld-cli --config

Once configured, simply type aworld-cli in your terminal to start your journey.

Alternatively, you can configure by creating a .env file in your working directory with your model and API settings. See Environment configuration for details.

Automate Creation with AWorld-CLI

<p align="justify"> AWorld-CLI goes beyond simple scaffolding. It acts as a central brain, the AWorld Agent, which orchestrates a team of specialized sub-agents to build, evaluate, and even evolve other agents autonomously.

This multi-agent system works in concert to turn your ideas into reality:

</p> <table> <thead> <tr><th style="white-space:nowrap">Agent Name</th><th>Role & Core Function</th></tr> </thead> <tbody> <tr><td style="white-space:nowrap">👑 AWorld Agent</td><td><strong>The Orchestrator</strong>: The central brain that interprets user goals, creates a plan, and delegates tasks to the appropriate sub-agents. It manages the entire workflow from start to finish.</td></tr> <tr><td style="white-space:nowrap">🧑‍💻 Developer</td><td><strong>The Builder</strong>: The master craftsman responsible for writing, debugging, and refactoring code.</td></tr> <tr><td style="white-space:nowrap">🧐 Evaluator</td><td><strong>The Judge</strong>: The quality assurance expert. It assesses the Developer's output against objective criteria, providing the critical feedback required for the evolution loop.</td></tr> </tbody> </table>

The Evolution Loop: Build -> Evaluate -> Evolve

Imagine you ask: "Help me create an English word learning mini-app with a UI quality score above 0.9."

  • The Developer Builds: The Developer analyzes requirements and writes code (e.g., HTML) using CAST.
  • The Evaluator Judges: The Evaluator inspects the output using our verified Skill.
  • The Loop Refines: If the score is below target (e.g., 0.9), AWorld instructs the Developer to fix specific issues identified by the Evaluator. This loop continues until your criteria are met.

📹 See the Self-Evolution Loop in Action

<p align="center"> <video src="https://github.com/user-attachments/assets/ff56195e-e117-4d33-b709-9a2144680abd" poster="readme_assets/evolution_loop_poster.png" width="80%" controls style="max-width: 80%;"> </video> </p>

No Evaluation, No Evolution

<p align="justify"> For an agent to improve, it must first understand what "good" looks like. This evaluation is the core of our autonomous evolution loop, but it's a complex challenge. It ranges from <b>objective</b> tasks with clear metrics (e.g., solving a math problem) to <b>subjective</b> ones requiring human preference. Real-world evolution is further complicated by massive codebases, limited context windows, and the need for precise iteration. </p> <p align="justify"> AWorld provides the complete infrastructure to master both evaluation scenarios, turning your expertise into the definitive driving force that steers an agent through the entire evolution loop. </p>

CAST: Conquering Code Complexity

<p align="justify"> Agents often fail because of overwhelming code complexity. We built <b>CAST</b> (Code Abstract Syntax Tree) to solve this. Instead of seeing a flat text file, CAST gives the agent an architectural blueprint of the code. This enables: </p>
  • Hierarchical Navigation: Instantly understand code structure and purpose without getting lost in implementation details.
  • Nearly Infinite Context: Intelligently compresses code to feed the agent only relevant information, breaking the context window limitation.
  • Surgical Code Modification: Perform precise changes with full dependency awareness, avoiding the clumsy errors of "blind" text replacement.

Your Expertise as the Evaluator

<p align="justify"> CAST provides the technical capability for change, but your knowledge provides the direction. AWorld's <b>Shared Skill System</b> makes your expertise the ultimate measure of quality. </p> <p align="justify"> <b>Automated Evaluation</b>: <code>Evaluator</code> agent judge performance and identify flaws, setting a clear, objective target for the <code>Developer</code> agent. This creates a powerful synergy: the <code>Evaluator</code> sets the target, and the <code>Developer</code> uses the same knowledge to hit it. </p> <p align="justify"> <b>Human Evaluation</b>: For tasks demanding subjective judgment, your intuition is the ceiling. You are the ultimate evaluator. Provide natural language feedback at any stage, and the AWorld agent will interpret it as a high-priority instruction for the next evolutionary cycle. </p> <p align="justify"> Whether it's an automated score from a Skill you contributed or your direct manual guidance, in AWorld, precise feedback drives precise evolution. </p>

A Proven Harness: Benchmark Excellence

<p align="justify"> The following top rankings on competitive benchmarks are more than just agent achievements—they are direct validation of the AWorld **Harness**. They prove our robust, battle-tested infrastructure provides the essential foundation for building state-of-the-art AI.

Agent Benchmarking

<table style="width: 100%; border-collapse: collapse; table-layout: fixed;"> <thead> <tr> <th style="width: 30%; text-align: left; border-bottom: 2px solid #ddd; padding: 8px;">Category</th> <th style="width: 20%; text-align: left; border-bottom: 2px solid #ddd; padding: 8px;">Achievement</th> <th style="width: 20%; text-align: left; border-bottom: 2px solid #ddd; padding: 8px;">Performance</th> <th style="width: 25%; text-align: left; border-bottom: 2px solid #ddd; padding: 8px;">Key Innovation</th> <th style="width: 5%; text-align: left; border-bottom: 2px solid #ddd; padding: 8px;">Date</th> </tr> </thead> <tbody> <tr> <td style="padding: 8px; vertical-align: top;">🤖 Agent <br> <a href="https://playground.aworldagents.com/" target="_blank" style="text-decoration: none;"> <img src="https://img.shields.io/badge/Try-Online-9B59B6?style=flat-square" alt="Try Online"> </a> </td> <td style="padding: 8px; vertical-align: top;"> <strong>GAIA Benchmark <br>Excellence</strong> <br> <a href="https://huggingface.co/spaces/gaia-benchmark/leaderboard" target="_blank" style="text-decoration: none;"> <img src="https://img.shields.io/badge/GAIA-Leaderboard-blue" alt="GAIA"
View on GitHub
GitHub Stars1.1k
CategoryEducation
Updated1d ago
Forks119

Languages

Python

Security Score

100/100

Audited on Mar 20, 2026

No findings