The paper introduces skilder, a framework that organizes LLM agent capabilities into role‑scoped bundles of skills, tools, and instructions, with explicit limits. Agents start with a minimal role catalog, discover the roles needed for a task, and receive the associated tools only through a single MCP server, ensuring deterministic enforcement of scope. Experiments on 13 tasks with six models show that skilder’s authorization layer prevents unauthorized tool calls and parameter violations while maintaining flexibility through dynamic cross‑role capability acquisition.
arXiv:2602. 12430v4 Announce Type: replace-cross Abstract: The transition from monolithic language models to modular, skill-equipped agents marks a defining shift in how large language models (LLMs) are deployed in practice.
ClawSentry is an open‑source, framework‑agnostic security supervision gateway designed to protect autonomous large language model (LLM) agents from progressive risks that can arise at four points in the agent control loop: skill admission, invocation‑time intent, execution‑time effect, and post‑action consequence. It introduces a multi‑tier decision engine—deterministic L1, rule‑anchored L2, and read‑only L3—alongside a First‑Use Skill Package Review (FSPR) and an Agent Harness Protocol (AHP) that applies a single policy across multiple agent runtimes without modifying their internals. Evaluation on SkillInject and the SkillsSafety benchmark shows that ClawSentry significantly reduces contextual adversarial skill risk (ASR) while maintaining high task success rates (TSR).
The paper introduces a post‑training framework that teaches a 4B‑parameter language model to exercise task‑conditioned authority in executable terminal and Model Context Protocol (MCP) environments. By auditing each action across six risk dimensions with deterministic verifiers and optimizing for task‑specific excess‑privilege values, the authors achieve 98.48% safe success and reduce excess‑authority errors from 4.56% to 0.79% on held‑out tasks. The study also demonstrates capability retention, prompt‑directed improvement, and generalization over a 400‑task continuation test.
The paper introduces AgentLeak, a black‑box attack that clones the task‑solving capabilities of a strong LLM agent onto a weaker one by exploiting differences between successful and failed executions. Unlike prior skill‑stealing methods that only recover explicit skill artifacts, AgentLeak identifies and incorporates missing procedural behaviors, boosting task pass rates by over 40% and closing more than 80% of the capability gap across 20 scenarios. The study demonstrates that observable execution behavior can leak proprietary procedural knowledge, posing a confidentiality risk for LLM agents.
arXiv:2605. 26542v2 Announce Type: replace-cross Abstract: Tool-using agents increasingly operate in open-ended deployment environments, where they compose file systems, web APIs, code interpreters, and enterprise services at runtime.
TRUSS is a framework that generates and verifies automated agent skills, ensuring they are both functionally effective and safe. It first checks functional claims against evidence and evaluates artifacts against nine safety properties, then tests admitted skills in a controlled environment to capture execution traces and identify failures. The approach achieves perfect precision and recall in vulnerability detection, significantly reduces attack success rates, and boosts task effectiveness and security rates in skill generation benchmarks.
TRUSS is a framework that generates and verifies automated agent skills, ensuring they are both functionally effective and safe. It evaluates candidate skills against source evidence and nine safety properties, then tests them in a controlled environment to capture execution traces and identify failures. The system iteratively refines skills based on these results, achieving high precision in vulnerability detection and significantly improving task performance and security rates.
The paper introduces Many-Tier Instruction Hierarchy (ManyIH), a new framework for resolving conflicts among instructions with arbitrarily many privilege levels in large language model agents. It presents ManyIH-Bench, a benchmark featuring 853 agentic tasks that require navigating up to 12 levels of conflicting instructions across 46 real-world agents. Experiments show current models achieve only about 40% accuracy when instruction conflict scales, highlighting a gap in fine-grained, scalable conflict resolution.
The paper evaluates a task-based permission scoping architecture for AI agents, comparing a fine‑tuned RoBERTa‑large encoder to few‑shot Claude Haiku 4.5 on a 600‑prompt dataset. It shows the new system achieves comparable macro‑F1 (0.881 vs. 0.886) and higher precision (0.897 vs. 0.842), while reducing severity‑weighted residual risk from 1.12 to 0.63. The study also introduces an attack‑surface elimination metric, demonstrating that task‑granular control can close 84.4% of the severity‑weighted surface, far surpassing role‑based ceilings alone.
arXiv:2608.30041v1 Announce Type: cross Abstract: Large language model agents place outputs from external skills into their execution context, allowing attacker-controlled data to influence later pri...
arXiv:2607. 03423v1 Announce Type: cross Abstract: Modern AI agent implementations such as frontier coding agents chain multiple tools at runtime that create a security surface that per-tool guardrails are unable to address, as individually permitted tools can violate organizational policies when composed.
arXiv:2605. 18414v2 Announce Type: replace-cross Abstract: Large language models increasingly operate as autonomous agents that select and invoke tools from large registries.