AI Paper Atlas

Research map / Multi-agent systems and games / Human-AI and goal-conditioned

Zero-Shot Coordination among LLM Agents

Adrian Hayler, Shashank Reddy Chirra, Andrei Lupu, Johannes Forkel, Bidipta Sarkar, Siheng Feng, Jakob Foerster

NeurIPS 2026 · Multi-agent systems and games › Human-AI and goal-conditioned · approach: Collaborative multi-agent perception · 0 citations (OpenAlex)

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.

Studies zero-shot coordination between independently developed LLM agents, without shared specialised scaffolding, at test time.

Problem

Zero-shot coordination (ZSC) of independently developed LLM agents at test time; existing LLM coordination work often relies on specialised scaffolding that independent agents are unlikely to share.

Approach

Evaluates LLM agents in ZSC settings without assuming shared specialised scaffolding, contrasting with prior LLM coordination work that depends on such scaffolding. Abstract is truncated, so the specific method is not stated.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add hand-crafted Coordination Friendly Definition (CFD) to prompttext agents (Grok 4.1 Fast, GPT-5.2) · nine coordination environmentscoordination failure rate: reduced in 6/9 (Grok) and 5/9 (GPT-5.2) environmentsImproved
add hand-crafted CFD to prompttext agents · Cat & Dogjoint reward: consistent safe fallback reward of 5Improved
add hand-crafted CFD to code agentscode agents · SP-XP gap across environmentsSP-XP gapNo clear effect
add hand-crafted CFDtext agents · ad-hoc conventionsconvention formationHurt
use meta-learned CFD instead of hand-craftedGrok 4.1 Fast text agent · self-play scorescoreImproved
use meta-learned CFD trained on code agents with a policy criticother models and generic code scaffold · XP_LLMXP score: modest gainsImproved
use meta-learned CFD across LLM typescode agents · XP_seedXP scoreNo clear effect
increase reasoning effort or model sizetext agents · Cat & Dogreveal rate / ZSC performanceMixed
add prompt stating partner is identical instance of the same modeltext agents · Cat & Dogreveal rate: GPT-5.2: ~75% -> ~5%Hurt
relabel signals or translate environment to Germantext agents · Cat & Doglocal policyMixed

Limitations

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