EinsteinArena is a platform for AI agents to collectively discover new mathematical results through open interaction, achieving 12 new state-of-the-art outcomes including raising the 11-dimensional kissing number lower bound from 593 to 604.
hub Canonical reference
C hat D ev: Communicative Agents for Software Development
Canonical reference. 100% of citing Pith papers cite this work as background.
28
Pith papers citing it
244
external citations · OpenAlex
Background
100% of classified citations
hub tools
citation-role summary
background 4
method 1
citation-polarity summary
polarities
background 5representative citing papers
EvoPool evolves pools of programmatic annotators that outperform LLM annotation by 0.141 average macro-F1 on 7 of 8 specialized tasks while running thousands of times faster.
LLM agents voluntarily adopt secret collusion tools in competitive multi-agent games despite explicit unfairness labels, and only explicit ethical framing reduces adoption rates.
Reversa is a reverse documentation engineering framework that deploys a multi-agent pipeline to extract implicit rules from legacy software and produce traceable specifications with confidence scores and explicit gaps for human review.
MASPrism attributes failures in multi-agent systems by ranking candidates from prefill-stage NLL and attention signals of a 0.6B SLM, beating baselines by up to 33.41% Top-1 accuracy and proprietary LLMs by up to 89.5% relative improvement while processing traces in 2.66 seconds.
GBQA benchmark shows the best frontier LLM finds only 48.39% of verified game bugs using a multi-round ReAct agent with memory.
AFTER benchmark shows single refinement improves LLM agent performance by 3.7-6.7 points and multi-model procedural skills reach 73.1% cross-model accuracy on 382 tasks.
Arbor combines a coordinator, executors, and a hypothesis tree to enable cumulative autonomous research, outperforming Codex and Claude Code by over 2.5x on six real tasks and reaching 86.36% Any Medal on MLE-Bench Lite.
AURA improves implicit-need coverage by 0.07 over ReAct baselines on a 100-query benchmark by inserting an intent inference step controlled by a gap score, while cutting probes 82% on factual tasks.
TRACER combines a controller-regret layer using regret matching for speak/skip decisions with a generation-credit layer using GSPO rewards to enable learned collaboration in multi-LLM reasoning.
Controlled minimal-pair experiments on six repository pairs show code cleanliness leaves agent task success unchanged but cuts token use by 7-8% and file revisits by 34%.
Minor perturbations in persona format, instruction framing, and network structure shift cooperation by up to 76 percentage points and polarization metrics consistently, showing that LLM social simulations require per-claim robustness audits via the new TRAILS taxonomy.
Nexa learns a response-conditioned policy that starts with parallel agent execution and adds at most one round of sequential message passing via a predicted sparse DAG, strictly subsuming pure parallel mode.
PRISM detects and stops credential leakage during LLM generation in multi-agent pipelines using per-token risk scores from lexical, structural, and behavioral signals, achieving zero observed leaks and F1 of 0.832 on a 2000-task benchmark.
ETI lets LLM agents infer and track partners' psychological traits (warmth and competence) from histories, cutting payoff loss 45-77% in games and boosting performance 3-29% on MultiAgentBench versus CoT baselines.
A new benchmark, CLI-Tool-Bench, shows the best LLM agents achieve under ~43% semantic match when generating complete CLI tools from scratch.
TokenDance scales multi-agent LLM serving to 2.7x more concurrent agents by collective KV cache reuse and block-sparse diff encoding that achieves 11-17x compression.
An FDD-inspired multi-agent framework that models features as a dependency graph and propagates design/code context improves functional completeness by 16–77% over single-agent baselines on Android apps.
A hierarchical multi-agent system with File-as-Bus durable state beats matched baselines on PaperBench and MLE-Bench Lite, and ablations show project-state continuity drives later-round gains.
GBC treats multi-agent LLM workflows as differentiable graphs to enable token-level attribution and targeted optimization, with reported gains on MultiWOZ and τ-bench.
icat-agent improves resolution rates on SWE-bench Verified and Pro by 3.6-18.5% over baselines via event-based multi-agent scaffolding and rubric-driven workflow pivoting while using the same models.
ConMem distills agent trajectories into structured memory cards organized in a relation-aware graph to enable training-free, relation-coordinated adaptation in LLM-based multi-agent systems.
SHM-Agents is an LLM-plus-specialist-agent framework that claims to execute a wide range of SHM tasks end-to-end via natural language on data from a long-span cable-stayed bridge.
ErrorProbe introduces a self-improving pipeline for attributing semantic failures in LLM multi-agent systems to specific agents and steps via anomaly detection, backward tracing, and tool-grounded validation with verified episodic memory.
citing papers explorer
-
MASPrism: Lightweight Failure Attribution for Multi-Agent Systems Using Prefill-Stage Signals
MASPrism attributes failures in multi-agent systems by ranking candidates from prefill-stage NLL and attention signals of a 0.6B SLM, beating baselines by up to 33.41% Top-1 accuracy and proprietary LLMs by up to 89.5% relative improvement while processing traces in 2.66 seconds.