CUBE: A Standard for Unifying Agent Benchmarks

TL;DR

CUBE is a universal standard built on MCP and Gym that unifies agent benchmarks, reducing integration effort and enabling seamless evaluation across platforms.

Contribution

It introduces CUBE, a standard protocol that separates concerns into API layers, facilitating universal access to benchmarks without custom integration.

Findings

CUBE enables wrapping benchmarks once for use across platforms.

It reduces integration time and effort for agent evaluation.

Community adoption of CUBE can prevent further fragmentation.

Abstract

The proliferation of agent benchmarks has created critical fragmentation that threatens research productivity. Each new benchmark requires substantial custom integration, creating an "integration tax" that limits comprehensive evaluation. We propose CUBE (Common Unified Benchmark Environments), a universal protocol standard built on MCP and Gym that allows benchmarks to be wrapped once and used everywhere. By separating task, benchmark, package, and registry concerns into distinct API layers, CUBE enables any compliant platform to access any compliant benchmark for evaluation, RL training, or data generation without custom integration. We call on the community to contribute to the development of this standard before platform-specific implementations deepen fragmentation as benchmark production accelerates through 2026.