SkillWeaver formalizes compositional skill routing for LLM agents and introduces SAD, which raises step-level decomposition accuracy from 51% to 67.7% on a new 300-query benchmark over 2209 real MCP skills.
hub Canonical reference
Taskweaver: A code-first agent framework
Canonical reference. 100% of citing Pith papers cite this work as background.
24
Pith papers citing it
4
external citations · external index
Background
100% of classified citations
hub tools
citation-role summary
background 8
citation-polarity summary
roles
background 8polarities
background 8representative citing papers
TimeSage-MT introduces a multi-turn benchmark for agentic time series reasoning and shows frontier LLMs drop sharply on decision-oriented tasks due to memory and uncertainty failures.
GraphFlow uses a unified wGraph to dynamically instantiate workflows and manage KV caches for LLM agents, reporting 4.95 pp average gains and 4x memory reduction on five benchmarks.
A new 7x4 taxonomy organizes agentic AI security threats by architectural layer and persistence timescale, revealing under-explored upper layers and missing defenses after surveying 116 papers.
MatClaw shows a code-first LLM agent autonomously generating and executing workflows for ML force field training, Curie temperature prediction, and parameter search on CuInP2S6, succeeding on code but requiring interventions for tacit domain knowledge.
Semi-structured interviews with 13 early adopters in one large tech organization identify multiple framings of transparency and synthesize them into a developer-user-governance multidimensional framework.
LiL vulnerabilities are more severe than ecosystem and conventional bugs and drop LLM-based repair Pass@1 by ~10.8%, with three categories often at 0% success.
Empirical analysis across 15 LLMs and 1,141 skills identifies a logarithmic routing decay law and a multiplicative execution law coupled by a single fitted slope parameter b that enables targeted library optimizations improving routing accuracy and downstream task pass rates.
STAR presents a failure-aware routing framework using a state-conditioned transition policy and an agent routing matrix combining expert routes with learned recoveries from execution traces to improve multi-agent spatiotemporal reasoning.
Claw-Eval-Live benchmark with 105 tasks shows no frontier LLM agent exceeds 66.7% success rate on evolving real-world workflows, with HR and multi-system tasks as persistent bottlenecks.
An empirical evaluation of 22 agentic frameworks on BBH, GSM8K, and ARC benchmarks shows stable performance in 12 frameworks but highlights orchestration failures and weaker mathematical reasoning.
Claw-Eval is a new trajectory-aware benchmark for LLM agents that records execution traces, audit logs, and environment snapshots to evaluate completion, safety, and robustness across 300 tasks, revealing that opaque grading misses 44% of safety issues.
The paper systematizes agentic skills beyond tool use, providing design pattern and representation-scope taxonomies plus security analysis of malicious skill infiltration in agent marketplaces.
AgentXRay formulates workflow reconstruction as combinatorial optimization and uses Monte Carlo Tree Search with Red-Black Pruning to approximate black-box agent behaviors via output-based proxy metrics.
LLM Orchestration integrates modality experts via an LLM controller, cross-modal memory, and interaction layer to enable multimodal input-output without gradient-based training.
OS-Atlas, trained on the largest open-source cross-platform GUI grounding corpus of 13 million elements, outperforms prior open-source models on six benchmarks across mobile, desktop, and web platforms.
DA-Studio is an agentic system that generates, executes, and exposes multi-step data analysis workflows in a sandboxed environment with visible traces and artifacts.
On a rebalanced WildChat 12-agent benchmark, fine-tuned encoders best match multi-agent gold sets, while Weighted Agent Routing improves cost-aware utility on strong scorers.
Bian Que is an agentic framework using a unified operational paradigm, flexible Skill Arrangement, and self-evolving mechanism to automate O&M tasks, achieving 75% alert reduction and over 50% MTTR cut in production deployment.
A two-step agentic system for extracting insights from VSM simulations achieves up to 86% accuracy with top LLMs by using progressive data discovery and slim context.
An MCP-native workflow engine decouples agent reasoning from execution by using declarative blueprints, reducing token cost by over 99% on a 67-step Kubernetes synchronization task.
The Dynamic Tiered AgentRunner framework uses risk-adaptive tiering, separation of powers across agents, and verifier-recovery loops to enable governable and resilient enterprise AI execution.
SDOF combines an RLHF-trained intent router with a state-aware dispatcher using finite automata to constrain multi-agent orchestration, reporting 80.9% routing accuracy and 86.5% task completion on a recruitment platform while blocking unsafe actions.
A survey consolidating frameworks, data practices, large action models, benchmarks, applications, and research gaps in LLM-brained GUI agents.
citing papers explorer
-
Compositional Skill Routing for LLM Agents: Decompose, Retrieve, and Compose
SkillWeaver formalizes compositional skill routing for LLM agents and introduces SAD, which raises step-level decomposition accuracy from 51% to 67.7% on a new 300-query benchmark over 2209 real MCP skills.
-
TimeSage-MT: A Multi-Turn Benchmark for Evaluating Agentic Time Series Reasoning
TimeSage-MT introduces a multi-turn benchmark for agentic time series reasoning and shows frontier LLMs drop sharply on decision-oriented tasks due to memory and uncertainty failures.
-
GraphFlow: A Graph-Based Workflow Management for Efficient LLM-Agent Serving
GraphFlow uses a unified wGraph to dynamically instantiate workflows and manage KV caches for LLM agents, reporting 4.95 pp average gains and 4x memory reduction on five benchmarks.
-
A Systematic Survey of Security Threats and Defenses in LLM-Based AI Agents: A Layered Attack Surface Framework
A new 7x4 taxonomy organizes agentic AI security threats by architectural layer and persistence timescale, revealing under-explored upper layers and missing defenses after surveying 116 papers.
-
MatClaw: An Autonomous Code-First LLM Agent for End-to-End Materials Exploration
MatClaw shows a code-first LLM agent autonomously generating and executing workflows for ML force field training, Curie temperature prediction, and parameter search on CuInP2S6, succeeding on code but requiring interventions for tacit domain knowledge.
-
"So There's a Catch-22 Here": How Early Adopters Who Build Multi-Agent LLM Systems Conceptualize Transparency
Semi-structured interviews with 13 early adopters in one large tech organization identify multiple framings of transparency and synthesize them into a developer-user-governance multidimensional framework.
-
Towards Demystifying and Repairing LLM-in-the-Loop Vulnerabilities
LiL vulnerabilities are more severe than ecosystem and conventional bugs and drop LLM-based repair Pass@1 by ~10.8%, with three categories often at 0% success.
-
The Scaling Laws of Skills in LLM Agent Systems
Empirical analysis across 15 LLMs and 1,141 skills identifies a logarithmic routing decay law and a multiplicative execution law coupled by a single fitted slope parameter b that enables targeted library optimizations improving routing accuracy and downstream task pass rates.
-
STAR: Failure-Aware Markovian Routing for Multi-Agent Spatiotemporal Reasoning
STAR presents a failure-aware routing framework using a state-conditioned transition policy and an agent routing matrix combining expert routes with learned recoveries from execution traces to improve multi-agent spatiotemporal reasoning.
-
Claw-Eval-Live: A Live Agent Benchmark for Evolving Real-World Workflows
Claw-Eval-Live benchmark with 105 tasks shows no frontier LLM agent exceeds 66.7% success rate on evolving real-world workflows, with HR and multi-system tasks as persistent bottlenecks.
-
Agentic Frameworks for Reasoning Tasks: An Empirical Study
An empirical evaluation of 22 agentic frameworks on BBH, GSM8K, and ARC benchmarks shows stable performance in 12 frameworks but highlights orchestration failures and weaker mathematical reasoning.
-
Claw-Eval: Towards Trustworthy Evaluation of Autonomous Agents
Claw-Eval is a new trajectory-aware benchmark for LLM agents that records execution traces, audit logs, and environment snapshots to evaluate completion, safety, and robustness across 300 tasks, revealing that opaque grading misses 44% of safety issues.
-
SoK: Agentic Skills -- Beyond Tool Use in LLM Agents
The paper systematizes agentic skills beyond tool use, providing design pattern and representation-scope taxonomies plus security analysis of malicious skill infiltration in agent marketplaces.
-
AgentXRay: White-Boxing Agentic Systems via Workflow Reconstruction
AgentXRay formulates workflow reconstruction as combinatorial optimization and uses Monte Carlo Tree Search with Red-Black Pruning to approximate black-box agent behaviors via output-based proxy metrics.
-
Training-Free Multimodal Large Language Model Orchestration
LLM Orchestration integrates modality experts via an LLM controller, cross-modal memory, and interaction layer to enable multimodal input-output without gradient-based training.
-
OS-ATLAS: A Foundation Action Model for Generalist GUI Agents
OS-Atlas, trained on the largest open-source cross-platform GUI grounding corpus of 13 million elements, outperforms prior open-source models on six benchmarks across mobile, desktop, and web platforms.
-
DA-Studio: An Agentic System for End-to-End Data Analysis
DA-Studio is an agentic system that generates, executes, and exposes multi-step data analysis workflows in a sandboxed environment with visible traces and artifacts.
-
Multi-Agent Routing as Set-Valued Prediction: A WildChat Benchmark and Cost-Aware Evaluation
On a rebalanced WildChat 12-agent benchmark, fine-tuned encoders best match multi-agent gold sets, while Weighted Agent Routing improves cost-aware utility on strong scorers.
-
Bian Que: An Agentic Framework with Flexible Skill Arrangement for Online System Operations
Bian Que is an agentic framework using a unified operational paradigm, flexible Skill Arrangement, and self-evolving mechanism to automate O&M tasks, achieving 75% alert reduction and over 50% MTTR cut in production deployment.
-
Agentic Insight Generation in VSM Simulations
A two-step agentic system for extracting insights from VSM simulations achieves up to 86% accuracy with top LLMs by using progressive data discovery and slim context.
-
Separating Intelligence from Execution: A Workflow Engine for the Model Context Protocol
An MCP-native workflow engine decouples agent reasoning from execution by using declarative blueprints, reducing token cost by over 99% on a 67-step Kubernetes synchronization task.
-
Beyond Autonomy: A Dynamic Tiered AgentRunner Framework for Governable and Resilient Enterprise AI Execution
The Dynamic Tiered AgentRunner framework uses risk-adaptive tiering, separation of powers across agents, and verifier-recovery loops to enable governable and resilient enterprise AI execution.
-
SDOF: Taming the Alignment Tax in Multi-Agent Orchestration with State-Constrained Dispatch
SDOF combines an RLHF-trained intent router with a state-aware dispatcher using finite automata to constrain multi-agent orchestration, reporting 80.9% routing accuracy and 86.5% task completion on a recruitment platform while blocking unsafe actions.
-
Large Language Model-Brained GUI Agents: A Survey
A survey consolidating frameworks, data practices, large action models, benchmarks, applications, and research gaps in LLM-brained GUI agents.