AI Paper Atlas

Research map / Multi-agent systems and games / LLM agents

AgentSuite: Toward More Reliable Agent Evaluation with a Component-Based Benchmark Auditing Pipeline

Hyewon Suh, Binfei Ji, Seojune Lee, Rishi Khare, Basit Khan, Hyunjun Kim, Tianyi Zhang, Venkat Krishna Srinivasan, Peter Belcak, Shizhe Diao, Pavlo Molchanov, Yingyan Celine Lin and 1 more

ICML 2026 · Multi-agent systems and games › LLM agents · approach: Agentic benchmarks and orchestration

Read the paperCheck your idea against 137,217 papers

AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Proposes COBA, a component-based auditing pipeline that checks agent benchmarks for hidden flaws across instructions, environments, tools, ground-truth trajectories, and evaluation protocols.

Problem

LLM agent benchmarks often contain hidden flaws from interactions among instructions, environments, tools, ground-truth trajectories, and evaluation protocols, which confound model errors with benchmark artifacts and undermine leaderboard comparisons.

Approach

A component-based auditing pipeline (COBA) that decomposes an agent benchmark into its constituent components and audits each, including their interactions, to surface benchmark validity flaws before trusting agent evaluation results.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use taxonomy-guided prompt instead of generic critique prompt (ablation)COBA judge · ACEBench disagreement caseswin rate on disagreements: 72.8%Improved
use taxonomy-guided prompt instead of generic critique prompt (ablation)COBA judge · CFB (ComplexFuncBench) disagreement caseswin rate on disagreements: 73.3%Improved
use taxonomy-guided prompt instead of generic critique prompt (ablation)COBA judge · ACEBench and CFBnumber of flagged issues: generic flags 1.7x more issuesImproved
remove flawed tasks via COBA filtering vs random removal of same numberACEBench ranking · ACEBenchKendall tau: 0.871 vs 0.960Improved
remove flawed tasks via COBA filtering vs random removal of same numberACEBench ranking · ACEBenchpairwise flip rate: 5.75% vs 1.50%Improved
remove inflation-prone flawed tasksACEBench · ACEBenchaverage model performance: 1.000 -> 0.780Improved
remove deflation-dominated flawed tasksACEBench · ACEBenchaverage model performance: 0.555 -> 0.780Improved
filter flagged tasks before evaluationgpt-4o-20240806 · ACEBenchscore: 85.2 -> 88.3Improved
filter flagged tasks before evaluationgpt-oss-120b · ACEBenchrank: 11 -> 16Hurt
filter flagged tasks before evaluationclaude-4.5-sonnet-thinking-on-10k · ACEBenchrank: 6 -> 3Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy