Back to directory

AWorld

An agent harness and platform for building autonomous AI agents that orchestrate tools, memory and context, letting experts encode domain knowledge into reusable skills.

AWorld

AWorld describes itself as a harness for building autonomous agents: a framework that orchestrates an agent's tools, memory, context and execution rather than relying on a model alone. It aims to let a domain expert embed their own workflows and intuition into that harness, turning expert knowledge into a fleet of agents that can act in a specific domain.

The project demonstrates this by encoding expertise into reusable skills that other agents can call on, with examples including automatic app creation evaluated by a separate UI-evaluation skill, deep-search research workflows carried out by an agent-browser skill, and one-prompt video generation through Remotion or video-diffusion skills.

AWorld is written in Python and released under the MIT license.

More in Agent frameworks

See all 902
  • Recurse

    recurse.run

    A serverless harness that lets a coding agent design, test, and deploy its own specialist sub-agents as tools, MCP servers, or bots, defined by a schema-validated manifest.

  • ProofForge

    shulgin.is-a.dev

    An AI-agent pipeline that decomposes a mathematics problem, proves each piece, and formalizes the result in Lean 4 and Mathlib until the Lean kernel accepts it, with six of its proofs already merged into Google DeepMind's formal-conjectures repository.

  • local-coder

    github.com

    Builds a local coding-agent team on top of Ollama and OpenCode by detecting the machine's hardware, recommending models for seven specialist roles, and testing whether each assigned model can actually use tools, edit a repository and pass verification.