AppWorld

AppWorld is a benchmark and simulated environment that tests how well coding agents use app APIs for developers building and evaluating agents.

What it is good for

  • Testing how well an AI agent calls app functions and APIs
  • Comparing coding agents on realistic multi-step tasks
  • Running agent experiments against a simulated set of apps
  • Connecting a terminal coding agent to simulated apps through MCP

At a glance

Who it is for
Setup
Requires programming

Level 5 of 5. You need to write code to use it.

Cost
Free and open source
Type
Framework or SDK
Tasks
Works with
Codex, Gemini, Claude Desktop, Docker

Momentum

Rank this week
Not ranked yet
Best rank
None yet
GitHub stars
544

GitHub numbers as of Nov 2, 2026. Updated weekly.

516 531 546 Sep 24, 2026 Nov 2, 2026
GitHub stars for AppWorld: 518 on Sep 24, 2026, 544 on Nov 2, 2026. Source: GitHub.

Official links

Setup steps are on the official pages. We do not repeat them here, since they change.

Questions

Is AppWorld free?
Yes. It is free and open source (Apache-2.0 license).
Do I need to know how to code to use AppWorld?
Yes. You need to write code to use it.
What does AppWorld work with?
According to its own documentation, it works with Codex, Gemini, Claude Desktop, and Docker.
What is AppWorld good for?
Testing how well an AI agent calls app functions and APIs, comparing coding agents on realistic multi-step tasks, running agent experiments against a simulated set of apps, and connecting a terminal coding agent to simulated apps through MCP.

Similar agents

  • OpenSandbox

    OpenSandbox is a sandbox runtime that lets developers run AI agents in isolated environments on their own infrastructure.

  • LangChain

    LangChain is a framework that helps developers build AI agents and LLM-powered applications that connect to models and data sources.

  • crewAI

    CrewAI is a Python framework that lets developers build teams of AI agents that split up and complete multi-step work.

  • LangGraph

    LangGraph is a framework that helps developers build long-running agents that keep their state and resume after failures.

  • Agno

    Agno is a framework and service platform that builds, runs, and improves AI agents for developers and teams who need them available as an API or MCP server.

  • AgentScope

    AgentScope is a framework that helps developers build and run AI agents they can monitor and trust.

Last updated Nov 2, 2026. Tracked since Oct 29, 2026. The summary was written by an AI model from the project's own description and checked against our copy rules. Spotted an error? Tell us.