Papers

Field

Top tools here

Coming with more sources: stars history

All tools

Agno4 releases this week
FastMCP6 releases this week
n8n+25% downloads · 148.3k/wk+25.0%
DeepEval+15% downloads · 769.3k/wk+14.7%
Ragas+14% downloads · 319.9k/wk+13.6%

Top tools here

Open problems in Agents

Long-horizon task execution and evidence accumulation3 papers raise it · 2 address it
Agents struggle to sustain multi-step reasoning, adapt strategies, and turn intermediate results into final deliverables in real-world tasks.
- Scaling Agentic Data for Long-Horizon Terminal Intelligenceraises · addresses
- FDE-Gym: The Last Mile of AI Deliveryraises · addresses
- AgentSWE: Can Coding Agents Build the Agent You Actually Want?raises
Realistic evaluation of agentic learning from experience1 paper raises it · 1 addresses it
Existing benchmarks often test reasoning with pre-existing knowledge rather than learning from novel, unfamiliar environments.
- Learn2Play Bench: How Well Do LLM Agents Learn from Experience in Unfamiliar Environments?raises · addresses
Faithful agent development from natural-language requirements1 paper raises it · 1 addresses it
Coding agents often fail to deliver agents that truly meet the specified requirements, not just appear to.
- AgentSWE: Can Coding Agents Build the Agent You Actually Want?raises · addresses
Detecting when past context matters in long conversations1 paper raises it · 0 address it
AI systems cannot reliably identify when historical context is relevant for current interactions in long-term companionship.
- RealCompanion: Benchmarking Human Understanding from Reasoning over Longitudinal Real-World Conversationsraises
Scalable and realistic GUI agent training and evaluation1 paper raises it · 2 address it
High cost and scarcity of realistic interactive environments limit progress in GUI agent training and evaluation.
- MAI-Gym: An Open All-in-One Platform for GUI Agent Evaluation and RL Training across Mobile and Computer Environmentsraises · addresses
- OpenGCUA: Toward a Fully Open Stack for Generalist Computer Useaddresses

Named by an AI from these papers' abstracts.

Open problems

Recently funded

Market

Asymptote LabsSeed · Oct 20$5.5M
GreptileSeries B+ · Oct 19$50M
VesslaPre-seed · Oct 19$3.3M
SupaHumanSeries A · Oct 19$1.5M
BoxLiteSeed · Oct 19undisclosed

Recently funded

Topics in this field

coding agents · code generation · coding agent286 papers77 this week
skills · tasks · ai308 papers60 this week
ai agents · indirect prompt injection · adversarial314 papers45 this week
recursive self improvement · harness evolution · adaptive260 papers45 this week
tool using agents · agent systems · multi agent209 papers45 this week
vision language · safety evaluations · safety alignment115 papers37 this week

Topics in this field

Active contributors · 90 days

Active contributors

Activity · 12 weeks

Coming with more sources: repos

Activity

Recent papers

All in Research →

arXiv cs.AIOct 17
MELD: Recursive Self-Improvement via Hierarchical Coordination for Multi-Agent Scientific Discovery
Sizhe Tang, Zuyuan Zhang, Zeyu Fang, Tian Lan
Multi-agent scientific discovery with large language models raises a coupled learning problem: research assignments determine both which experiments are pursued and the experiences from which agents adapt. We introduce MELD, a unified hierarchical test-time training framework that jointly adapts a team-level Coordinator policy and agent-level individual Scientist policies within a single discovery process. The Coordinator generates a joint portfolio of search branches and research directives over a shared derivation tree, while Scientists develop hypotheses and execution plans implemented by a shared frozen Executor. Both levels receive alternating GRPO-style parameter updates under a shared smooth best-of-budget objective, allowing team organization and individual research behavior to co-evolve in a recursive self-improvement loop. Our analysis establishes objective alignment for the hierarchy, characterizes conditions for coordination gains over independent parallel discovery and the first-order benefit of private adaptation, and derives a finite-budget regret bound for adapted models. On 22 MLGym and MLE-bench tasks, MELD is best on 20 of them and achieves the highest…
PDF arXiv
arXiv cs.AIOct 17
IF-MAP: Influence-Function-Inspired Agent Prioritization in Multi-Agent Systems
Jie Mao, Pingchen Lu, Zikun Qu, Jiahang Sun, Xiangyi Wang +4
Improving every agent in an LLM-based multi-agent system can be prohibitively costly, making it important to identify which agent should be prioritized to achieve the greatest system-level gain with limited resources. Failure attribution can inform this decision, but an agent's responsibility for failures does not directly reveal the system-level gains achievable by improving it. We therefore introduce IF-MAP, an influence-function-inspired method for identifying agents with high potential for system-level improvement. For each agent, IF-MAP measures end-to-end loss changes under controlled semantic prompt perturbations, fits a regularized local response model in a low-dimensional prompt-embedding subspace, and uses the estimated response norm as a priority score. This black-box procedure requires no access to model parameters or gradients and produces agent priorities that can guide subsequent improvement effort. We evaluate IF-MAP across six multi-agent workflows spanning code generation, question answering, and mathematical reasoning, using the same priorities for prompt optimization and model upgrades. Optimizing the selected agents' prompts achieves a macro-average gain of…
PDF arXiv
arXiv cs.CROct 17
Defending Against Dormant Poisoning Attacks Across Language and Multimodal Agents
Wonjun Lee, Vaidehi Patil, Zaid Khan, Yue Zhang, Hyunji Lee +2
Recent work shows that attackers can implant dormant malicious behaviors that remain hidden at release but emerge after benign fine-tuning. As foundation models are used as agents, whether dormant poisoning extends to harmful agent actions and multimodal settings remains unclear. We show two agentic risks: (1) harmful behaviors activated by benign fine-tuning can extend to unseen actions; (2) multiple harmful behaviors can coexist and be jointly activated by benign downstream fine-tuning. Existing defenses often rely on external safe responses, which are hard to obtain for agent actions, or a trusted clean reference model. We exploit a key property of dormant poisoning: the attacker must preserve safe behavior at release to conceal poisoning, and this behavior can serve as a safety reference. Leveraging this property, we automatically construct a Safety Buffer pairing harmful inputs with the released model's responses and benign versions generated by the same model to remove harmful intent. The Safety Buffer can replace external safe responses in existing defenses, but these defenses do not distinguish updates that weaken safety from those that preserve it, often increasing…
PDF arXiv
arXiv cs.AIOct 17
Beyond Next-Screen Prediction: The Decision Value of GUI World Models
Sizhe Tang, Hanhan Zhou, Jeffrey Svajlenko, Rashmi Gangadharaiah
GUI world models are increasingly used to improve action selection by predicting the consequences of candidate actions, but it is unclear how much of the resulting gain comes from prediction itself rather than from the evaluator that interprets those predictions. We study this attribution problem through controlled interventions on GUI action selection. With candidates and predictions held fixed, lookahead gains can change with how the prediction is presented to the evaluator, so end-to-end gains cannot be attributed to prediction alone. We then introduce Direct Action Ranking (DAR), which learns from reference actions to score candidates without generating future screens, and compare it with a prediction-augmented variant under matched supervision. Predicted successors add no detectable ranking signal once evaluation is learned directly. Without retraining, DAR improves AndroidWorld task success by 5.8--16.4 points across three proposers, outperforms world-model lookahead, and transfers zero-shot to unseen apps, where every prediction-conditioned configuration falls below proposer confidence. These results suggest that GUI world models should be evaluated by their decision…
PDF arXiv
arXiv cs.LOOct 17
Tactix: Towards a Programmable Strategy Interface for Flexible SAT Solving
Haoze Wu, André Schidler, Armin Biere, Nina Narodytska
Modern SAT solvers combine a rich repertoire of proof rules with sophisticated strategies that determine when and how to apply them. Often, the internal implementations are tightly coupled to particular strategies, making it inconvenient to define new strategies without changing solver internals. We present Tactix, a proof-producing sequential CDCL solver that incorporates a wide range of techniques in state-of-the-art solvers with a clean separation of proof rules from solving strategies. This separation enables a public interface through which users can invoke particular techniques, observe their effects, and program a wide range of solving strategies. The default strategy of Tactix is programmable over this interface and outperforms state-of-the-art solvers on benchmarks from the 2024--2026 SAT Competitions. We also use the interface to develop novel strategies tailored to challenging benchmarks from two application domains, which led to performance gains over strong baselines. We intend Tactix to serve as an infrastructure for designing and evaluating novel solving strategies. Finally, we also present a self-study of the AI-assisted development process behind Tactix.
PDF arXiv
arXiv cs.LGOct 17
Measuring the Task Value of Precision Upgrades in Closed-Loop Agents
Sicheng Lyu, Xinyu Wang, Yu Gu, Lei Ding, Ziyang Song +1
A low-precision agent can call its full-precision (BF16) model at selected decisions as a fallback. A learned precision router chooses where to call it from the disagreement between the two models and is judged by end-to-end success. An inserted BF16 action, however, can rescue a failing task or damage a succeeding one. We argue that a fallback's value is the number of tasks it saves minus the number it spoils, measured for the intervention that will actually run. From restored states, we run paired branches that cross which model supplies the current action with which model continues, separating replacement of one action, takeover of later execution, and both. On Qwen3 agents with simulated 4-bit weights in tau2-bench, replacing the first action where the two models differed in tool use changed many task outcomes, but rescues and damages nearly cancelled, and on held-out telecom tasks the action-only estimate was negative. Replacing both the action and the continuation had a positive net gain in two of five exploratory lower-bit stress settings. Two references changed outcomes at fixed bit width: replacing an action with that of a second 4-bit checkpoint changed about as many…
PDF arXiv

Papers

Top tools here

Coming with more sources: stars history

All tools

Agno4 releases this week
FastMCP6 releases this week
n8n+25% downloads · 148.3k/wk+25.0%
DeepEval+15% downloads · 769.3k/wk+14.7%
Ragas+14% downloads · 319.9k/wk+13.6%

Topics in this field

coding agents · code generation · coding agent286 papers77 this week
skills · tasks · ai308 papers60 this week
ai agents · indirect prompt injection · adversarial314 papers45 this week
recursive self improvement · harness evolution · adaptive260 papers45 this week
tool using agents · agent systems · multi agent209 papers45 this week
vision language · safety evaluations · safety alignment115 papers37 this week

Recently funded

Market

Asymptote LabsSeed · Oct 20$5.5M
GreptileSeries B+ · Oct 19$50M
VesslaPre-seed · Oct 19$3.3M
SupaHumanSeries A · Oct 19$1.5M
BoxLiteSeed · Oct 19undisclosed

Active contributors · 90 days

Activity · 12 weeks

Coming with more sources: repos

Recent papers

All in Research →

arXiv cs.AIOct 17
MELD: Recursive Self-Improvement via Hierarchical Coordination for Multi-Agent Scientific Discovery
Sizhe Tang, Zuyuan Zhang, Zeyu Fang, Tian Lan
Multi-agent scientific discovery with large language models raises a coupled learning problem: research assignments determine both which experiments are pursued and the experiences from which agents adapt. We introduce MELD, a unified hierarchical test-time training framework that jointly adapts a team-level Coordinator policy and agent-level individual Scientist policies within a single discovery process. The Coordinator generates a joint portfolio of search branches and research directives over a shared derivation tree, while Scientists develop hypotheses and execution plans implemented by a shared frozen Executor. Both levels receive alternating GRPO-style parameter updates under a shared smooth best-of-budget objective, allowing team organization and individual research behavior to co-evolve in a recursive self-improvement loop. Our analysis establishes objective alignment for the hierarchy, characterizes conditions for coordination gains over independent parallel discovery and the first-order benefit of private adaptation, and derives a finite-budget regret bound for adapted models. On 22 MLGym and MLE-bench tasks, MELD is best on 20 of them and achieves the highest…
PDF arXiv
arXiv cs.AIOct 17
IF-MAP: Influence-Function-Inspired Agent Prioritization in Multi-Agent Systems
Jie Mao, Pingchen Lu, Zikun Qu, Jiahang Sun, Xiangyi Wang +4
Improving every agent in an LLM-based multi-agent system can be prohibitively costly, making it important to identify which agent should be prioritized to achieve the greatest system-level gain with limited resources. Failure attribution can inform this decision, but an agent's responsibility for failures does not directly reveal the system-level gains achievable by improving it. We therefore introduce IF-MAP, an influence-function-inspired method for identifying agents with high potential for system-level improvement. For each agent, IF-MAP measures end-to-end loss changes under controlled semantic prompt perturbations, fits a regularized local response model in a low-dimensional prompt-embedding subspace, and uses the estimated response norm as a priority score. This black-box procedure requires no access to model parameters or gradients and produces agent priorities that can guide subsequent improvement effort. We evaluate IF-MAP across six multi-agent workflows spanning code generation, question answering, and mathematical reasoning, using the same priorities for prompt optimization and model upgrades. Optimizing the selected agents' prompts achieves a macro-average gain of…
PDF arXiv
arXiv cs.CROct 17
Defending Against Dormant Poisoning Attacks Across Language and Multimodal Agents
Wonjun Lee, Vaidehi Patil, Zaid Khan, Yue Zhang, Hyunji Lee +2
Recent work shows that attackers can implant dormant malicious behaviors that remain hidden at release but emerge after benign fine-tuning. As foundation models are used as agents, whether dormant poisoning extends to harmful agent actions and multimodal settings remains unclear. We show two agentic risks: (1) harmful behaviors activated by benign fine-tuning can extend to unseen actions; (2) multiple harmful behaviors can coexist and be jointly activated by benign downstream fine-tuning. Existing defenses often rely on external safe responses, which are hard to obtain for agent actions, or a trusted clean reference model. We exploit a key property of dormant poisoning: the attacker must preserve safe behavior at release to conceal poisoning, and this behavior can serve as a safety reference. Leveraging this property, we automatically construct a Safety Buffer pairing harmful inputs with the released model's responses and benign versions generated by the same model to remove harmful intent. The Safety Buffer can replace external safe responses in existing defenses, but these defenses do not distinguish updates that weaken safety from those that preserve it, often increasing…
PDF arXiv
arXiv cs.AIOct 17
Beyond Next-Screen Prediction: The Decision Value of GUI World Models
Sizhe Tang, Hanhan Zhou, Jeffrey Svajlenko, Rashmi Gangadharaiah
GUI world models are increasingly used to improve action selection by predicting the consequences of candidate actions, but it is unclear how much of the resulting gain comes from prediction itself rather than from the evaluator that interprets those predictions. We study this attribution problem through controlled interventions on GUI action selection. With candidates and predictions held fixed, lookahead gains can change with how the prediction is presented to the evaluator, so end-to-end gains cannot be attributed to prediction alone. We then introduce Direct Action Ranking (DAR), which learns from reference actions to score candidates without generating future screens, and compare it with a prediction-augmented variant under matched supervision. Predicted successors add no detectable ranking signal once evaluation is learned directly. Without retraining, DAR improves AndroidWorld task success by 5.8--16.4 points across three proposers, outperforms world-model lookahead, and transfers zero-shot to unseen apps, where every prediction-conditioned configuration falls below proposer confidence. These results suggest that GUI world models should be evaluated by their decision…
PDF arXiv
arXiv cs.LOOct 17
Tactix: Towards a Programmable Strategy Interface for Flexible SAT Solving
Haoze Wu, André Schidler, Armin Biere, Nina Narodytska
Modern SAT solvers combine a rich repertoire of proof rules with sophisticated strategies that determine when and how to apply them. Often, the internal implementations are tightly coupled to particular strategies, making it inconvenient to define new strategies without changing solver internals. We present Tactix, a proof-producing sequential CDCL solver that incorporates a wide range of techniques in state-of-the-art solvers with a clean separation of proof rules from solving strategies. This separation enables a public interface through which users can invoke particular techniques, observe their effects, and program a wide range of solving strategies. The default strategy of Tactix is programmable over this interface and outperforms state-of-the-art solvers on benchmarks from the 2024--2026 SAT Competitions. We also use the interface to develop novel strategies tailored to challenging benchmarks from two application domains, which led to performance gains over strong baselines. We intend Tactix to serve as an infrastructure for designing and evaluating novel solving strategies. Finally, we also present a self-study of the AI-assisted development process behind Tactix.
PDF arXiv
arXiv cs.LGOct 17
Measuring the Task Value of Precision Upgrades in Closed-Loop Agents
Sicheng Lyu, Xinyu Wang, Yu Gu, Lei Ding, Ziyang Song +1
A low-precision agent can call its full-precision (BF16) model at selected decisions as a fallback. A learned precision router chooses where to call it from the disagreement between the two models and is judged by end-to-end success. An inserted BF16 action, however, can rescue a failing task or damage a succeeding one. We argue that a fallback's value is the number of tasks it saves minus the number it spoils, measured for the intervention that will actually run. From restored states, we run paired branches that cross which model supplies the current action with which model continues, separating replacement of one action, takeover of later execution, and both. On Qwen3 agents with simulated 4-bit weights in tau2-bench, replacing the first action where the two models differed in tool use changed many task outcomes, but rescues and damages nearly cancelled, and on held-out telecom tasks the action-only estimate was negative. Replacing both the action and the continuation had a positive net gain in two of five exploratory lower-bit stress settings. Two references changed outcomes at fixed bit width: replacing an action with that of a second 4-bit checkpoint changed about as many…
PDF arXiv

Fields · AI · updated 26m ago

Agents

LLM agents: planning, tool use, multi-agent systems, computer/web agents and agent benchmarks.

518papers this week−0.6% vs last week

All papers · 2.5k With code · 115

Newest Most relevant

2.4k papers this month

Open problems in Agents

Long-horizon task execution and evidence accumulation3 papers raise it · 2 address it
Agents struggle to sustain multi-step reasoning, adapt strategies, and turn intermediate results into final deliverables in real-world tasks.
- Scaling Agentic Data for Long-Horizon Terminal Intelligenceraises · addresses
- FDE-Gym: The Last Mile of AI Deliveryraises · addresses
- AgentSWE: Can Coding Agents Build the Agent You Actually Want?raises
Realistic evaluation of agentic learning from experience1 paper raises it · 1 addresses it
Existing benchmarks often test reasoning with pre-existing knowledge rather than learning from novel, unfamiliar environments.
- Learn2Play Bench: How Well Do LLM Agents Learn from Experience in Unfamiliar Environments?raises · addresses
Faithful agent development from natural-language requirements1 paper raises it · 1 addresses it
Coding agents often fail to deliver agents that truly meet the specified requirements, not just appear to.
- AgentSWE: Can Coding Agents Build the Agent You Actually Want?raises · addresses
Detecting when past context matters in long conversations1 paper raises it · 0 address it
AI systems cannot reliably identify when historical context is relevant for current interactions in long-term companionship.
- RealCompanion: Benchmarking Human Understanding from Reasoning over Longitudinal Real-World Conversationsraises
Scalable and realistic GUI agent training and evaluation1 paper raises it · 2 address it
High cost and scarcity of realistic interactive environments limit progress in GUI agent training and evaluation.
- MAI-Gym: An Open All-in-One Platform for GUI Agent Evaluation and RL Training across Mobile and Computer Environmentsraises · addresses
- OpenGCUA: Toward a Fully Open Stack for Generalist Computer Useaddresses

Named by an AI from these papers' abstracts.