SLBench: Evaluating How LLM Agents Follow Logical Relations in Skills

• URL: http://arxiv.org/abs/2607.09016v1
• Date: Fri, 10 Jul 2026 00:48:30 GMT
• Title: SLBench: Evaluating How LLM Agents Follow Logical Relations in Skills
• Abstract summary: We introduce SkillLogic, a framework for analyzing logical relations in skill files.<n>Using SkillLogic, we scan over 5000 public skills and find that 70% contain at least one logical relation.<n>We then construct SLBench, an 86-case executable benchmark from high-confidence, high-impact, and locally testable relations.
• Score: 15.716453891551057
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Agent skills extend LLM agents with reusable procedures, tools, and domain-specific workflows, but their safety depends on resolving dependencies among interacting instructions. We introduce SkillLogic, a framework for analyzing logical relations in skill files and constructing executable tests from them. Our taxonomy covers eight relation types, including preconditions that gate valid actions, constraints that limit how allowed actions may be performed, and fallbacks that specify recovery behavior after failure. Using SkillLogic, we scan over 5000 public skills and find that 70% contain at least one logical relation. We then construct SLBench, an 86-case executable benchmark from high-confidence, high-impact, and locally testable relations. Evaluating Codex and Claude Code across six LLM backbones shows unsafe rates up to 70%, with violations leading to privacy leaks, unsafe configuration changes, and incomplete cleanup. The human audit attributes failures to both agent capability gaps and low-salience skill text. We further show that SLGuard, a lightweight inference-time scaffold, reduces violations by 63% on targeted cases. Our results establish logical-relation following as a distinct reliability challenge for skill-guided agents.

Related papers

• Demystifying Agent Skills: Why They Work-Until They Don't [58.92875630998418]
  Existing evaluations largely measure whether skills improve aggregated task success, leaving a more fundamental question underexplored: emphtextbfWhen do skills help, why do they work, and where do they fail?<n>We isolate the effects of representation, outcome annotation, retrieval difficulty, and cross-framework of skills.<n>We consolidate these observations into a taxonomy of three high-level categories and twelve skill-use modes.
  arXiv  Detail & Related papers  (2026-08-14T07:26:38Z)
• REDAgentBench: Executable Red Teaming and Faithful Measurement of LLM Agent Systems [21.484325207774166]
  Adversarial inputs can exploit interactions between the agent and its environment, causing the agent to violate safety policies during execution.<n>We introduce REDAgentBench, an executable framework for autonomous red-teaming and faithful measurement.<n>It derives attacks from explicit safety constraints and associated agent-system vulnerabilities, runs them in isolated service sandboxes, and verifies harmful effects from service receipts and final-state changes.
  arXiv  Detail & Related papers  (2026-08-11T08:48:54Z)
• Cross-Layer Misalignment Detection in Agent Skills: A Progressive Loading-Aware Contrastive Learning Approach [12.79484512803839]
  Large language model (LLM) agents are increasingly extended through Agent Skills.<n>As open-source skill marketplaces expand, users and agents increasingly rely on brief metadata to select third-party skills.<n>We propose Progressive Loading-Aware Hierarchical Contrastive Learning (PL-HCL) to detect cross-layer misalignment.
  arXiv  Detail & Related papers  (2026-07-12T02:07:37Z)
• SkillVetBench: LLM-as-Judge for Multi-Dimensional Security Risk Evaluation in Open-Source LLM Agent Skills [0.0]
  We present SKILLVETBENCH, a live public leaderboard on Hugging Face that uses an LLM-as-Judge to vet agent skills.<n>S SARS (Skill Agentic Risk Score) is a five-dimensional agentic-risk metric with a principled weighted formula for instruction-following systems.
  arXiv  Detail & Related papers  (2026-06-14T16:30:33Z)
• SkillAudit: Ground-Truth-Free Skill Evolution via Paired Trajectory Auditing [81.51044612408793]
  SkillAudit is a framework for evolving agent skills without ground-truth feedback.<n>To turn behavioral differences into edit guidance, SkillAudit uses Process-Aligned Contrastive Evaluation.<n>It routes edits through two pipelines: Refine removes noisy or irrelevant guidance from broadly useful skills, while Repair replaces passages that conflict with the task.
  arXiv  Detail & Related papers  (2026-06-12T08:20:09Z)
• Behavioral Integrity Verification for AI Agent Skills [9.127363793428119]
  We formalize this as the behavioral integrity verification (BIV) problem.<n>On 49,943 skills from the OpenClaw registry, the deviation taxonomy reveals a pervasive description-implementation gap.<n>On a 906-skill malicious-skill detection benchmark, BIV reaches an F1 of 0.946, outperforming state-of-the-art rule-based and single-pass LLM baselines.
  arXiv  Detail & Related papers  (2026-05-12T08:41:09Z)
• ToolSafe: Enhancing Tool Invocation Safety of LLM-based agents via Proactive Step-level Guardrail and Feedback [53.2744585868162]
  Monitoring step-level tool invocation behaviors in real time is critical for agent deployment.<n>We first construct TS-Bench, a novel benchmark for step-level tool invocation safety detection in LLM agents.<n>We then develop a guardrail model, TS-Guard, using multi-task reinforcement learning.
  arXiv  Detail & Related papers  (2026-01-15T07:54:32Z)
• Towards Verifiably Safe Tool Use for LLM Agents [53.55621104327779]
  Large language model (LLM)-based AI agents extend capabilities by enabling access to tools such as data sources, APIs, search engines, code sandboxes, and even other agents.<n>LLMs may invoke unintended tool interactions and introduce risks, such as leaking sensitive data or overwriting critical records.<n>Current approaches to mitigate these risks, such as model-based safeguards, enhance agents' reliability but cannot guarantee system safety.
  arXiv  Detail & Related papers  (2026-01-12T21:31:38Z)
• Self-Abstraction from Grounded Experience for Plan-Guided Policy Refinement [61.35824395228412]
  Large language model (LLM) based agents are increasingly used to tackle software engineering tasks.<n>We propose Self-Abstraction from Grounded Experience (SAGE), a framework that enables agents to learn from their own task executions.
  arXiv  Detail & Related papers  (2025-11-08T08:49:38Z)
• SOPBench: Evaluating Language Agents at Following Standard Operating Procedures and Constraints [59.645885492637845]
  SOPBench is an evaluation pipeline that transforms each service-specific SOP code program into a directed graph of executable functions.<n>Our approach transforms each service-specific SOP code program into a directed graph of executable functions and requires agents to call these functions based on natural language SOP descriptions.<n>We evaluate 18 leading models, and results show the task is challenging even for top-tier models.
  arXiv  Detail & Related papers  (2025-03-11T17:53:02Z)
• Benchmarking LLMs and LLM-based Agents in Practical Vulnerability Detection for Code Repositories [8.583591493627276]
  We introduce JitVul, a vulnerability detection benchmark linking each function to its vulnerability-introducing and fixing commits.<n>We show that ReAct Agents, leveraging thought-action-observation and interprocedural context, perform better than LLMs in distinguishing vulnerable from benign code.
  arXiv  Detail & Related papers  (2025-03-05T15:22:24Z)
• Defining and Detecting the Defects of the Large Language Model-based Autonomous Agents [31.126001253902416]
  We present the first study focused on identifying and detecting defects in LLM Agents.<n>We collected and analyzed 6,854 relevant posts from StackOverflow to define 8 types of agent defects.<n>Our results show that Agentable achieved an overall accuracy of 88.79% and a recall rate of 91.03%.
  arXiv  Detail & Related papers  (2024-12-24T11:54:14Z)
• AgentBench: Evaluating LLMs as Agents [99.12825098528212]
  Large Language Model (LLM) as agents has been widely acknowledged recently.<n>We present AgentBench, a benchmark that consists of 8 distinct environments to assess LLM-as-Agent's reasoning and decision-making abilities.
  arXiv  Detail & Related papers  (2023-08-07T16:08:11Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.