AI Paper Atlas

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

Judgment Operators: A Composition-Invariant Substrate for Multi-Agent Action Spaces

Jun Li

ICML 2026 · Multi-agent systems and games › LLM agents · approach: Self-evolving language agent learning

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 judgment operators, a composition-invariant substrate for governing action spaces in heterogeneous multi-agent LLM systems, motivated by the failure of internal correction methods to enforce execution-time constraints.

Problem

Knowledge and governance fragment across agents in heterogeneous LLM multi-agent systems, causing composition-dependent behavior and violations that scale linearly with the number of agents.

Approach

A composition-invariant substrate called judgment operators that enforces execution-time constraints on agent action spaces, rather than relying on internal self-correction (Reflexion, CRITIC, Self-Refine), which the authors observe fails to enforce such constraints.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add precedent-based repair learning (JO-A over JO-static) (ablation)JO-static · Overcooked (N=200)Violation rate (%) and success (%): VR 2.6 -> 0.0; success 85.0 -> 100.0Improved
add external judgment operator JO-A to the agentNo Operator · Overcooked (N=200)Violation rate (%): 56.2 -> 0.0Improved
embed all constraints in a centralized system promptNo Operator · Overcooked (N=200)Violation rate / success (%): VR 56.2 -> 7.1; success 10.0 -> 2.0Mixed
add chain-of-thought constraint checkingNo Operator · Overcooked (N=200)Violation rate / success (%): VR 56.2 -> 14.8; success 10.0 -> 0.0Mixed
add Self-Refine internal correctionNo Operator · Overcooked (N=200)Violation rate / success (%): VR 56.2 -> 20.9; success 10.0 -> 0.0Mixed
add Reflexion internal self-correctionNo Operator · Overcooked (N=200)Violation rate (%): 56.2 -> 56.2No clear effect
add CRITIC internal correctionNo Operator · Overcooked (N=200)Violation rate (%): 56.2 -> 64.8Hurt
add LlamaGuard reject-only filteringNo Operator · Overcooked (N=200)Violation rate / success (%): VR 56.2 -> 67.8; success 10.0 -> 15.0Mixed
add JO-A mediation to agents (agent count change K1 to K2)Overcooked multi-agent · Composition sensitivity, OvercookedDelta comp (pp): 15.1pp -> 0.1ppImproved
add JO-A mediation to heterogeneous model mixProtocol agents · Composition sensitivity, ProtocolDelta comp (pp): 51.0pp -> 6.0ppImproved
increase bypass rate (partial mediation) from 0% to 30% (ablation)JO-A · Overcooked partial mediation (N=100)Violation rate (%): 0.0 -> 3.2 (bound 16.9)Hurt
apply JO-A with judgment artifacts learned by GPT-5.2 to a student model zero-shotGPT-4o-mini · WebArena-Wikipedia (N=200)Success (%): 2.4 -> 22.9Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Surprising findings

Limitations

Experimental setup

Baselines: Reflexion CRITIC Self-Refine

Builds on: Reflexion CRITIC Self-Refine

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