# STARS: Skill-Triggered Audit for Request-Conditioned Invocation Safety in Agent Systems

arXiv: https://arxiv.org/abs/2604.10286 · HTML: https://mlsanity.com/paper/2604.10286/

## Key facts

- Cited by 3 papers in the map (data through 14 Oct 2026).
- Placed in Agent threats, guardrails and governance, one of 16 directions in LLM agents and tool use.
- Builds on 16 of its 19 references that are in the map.
- First posted on arXiv 11 Apr 2026.

## Placed in

- LLM agents and tool use / [Agent threats, guardrails and governance](https://mlsanity.com/agents/agent-threats-guardrails-and-governance/)

## Builds on (most cited first, 16 of 16)

- [InjecAgent: Benchmarking Indirect Prompt Injections in Tool-Integrated Large Language Model Agents](https://mlsanity.com/paper/2403.02691/) (2024)
- [Agent-SafetyBench: Evaluating the Safety of LLM Agents](https://mlsanity.com/paper/2412.14470/) (2024)
- [Prompt Injection Attack to Tool Selection in LLM Agents](https://mlsanity.com/paper/2504.19793/) (2025)
- [Design Patterns for Securing LLM Agents against Prompt Injections](https://mlsanity.com/paper/2506.08837/) (2025)
- [RTBAS: Defending LLM Agents Against Prompt Injection and Privacy Leakage](https://mlsanity.com/paper/2502.08966/) (2025)
- [FalseReject: A Resource for Improving Contextual Safety and Mitigating Over-Refusals in LLMs via Structured Reasoning](https://mlsanity.com/paper/2505.08054/) (2025)
- [IPIGuard: A Novel Tool Dependency Graph-Based Defense Against Indirect Prompt Injection in LLM Agents](https://mlsanity.com/paper/2508.15310/) (2025)
- [Towards Verifiably Safe Tool Use for LLM Agents](https://mlsanity.com/paper/2601.08012/) (2026)
- [Unsafer in Many Turns: Benchmarking and Defending Multi-Turn Safety Risks in Tool-Using Agents](https://mlsanity.com/paper/2602.13379/) (2026)
- [Attractive Metadata Attack: Inducing LLM Agents to Invoke Malicious Tools](https://mlsanity.com/paper/2508.02110/) (2025)
- [AdapTools: Adaptive Tool-based Indirect Prompt Injection Attacks on Agentic LLMs](https://mlsanity.com/paper/2602.20720/) (2026)
- [AgenTRIM: Tool Risk Mitigation for Agentic AI](https://mlsanity.com/paper/2601.12449/) (2026)
- [Defense Against Indirect Prompt Injection via Tool Result Parsing](https://mlsanity.com/paper/2601.04795/) (2026)
- [Risky-Bench: Probing Agentic Safety Risks under Real-World Deployment](https://mlsanity.com/paper/2602.03100/) (2026)
- [AIR: Improving Agent Safety through Incident Response](https://mlsanity.com/paper/2602.11749/) (2026)
- [StepShield: When, Not Whether to Intervene on Rogue Agents](https://mlsanity.com/paper/2601.22136/) (2026)

## Built on by (most cited first, 3 of 3)

- [Safeguarding LLM Agents against Long-Horizon Threats via Shadow Memory](https://mlsanity.com/paper/2605.03228/) (2026)
- [Dynamic Agent Skills: A Lifecycle Survey and Taxonomy of Evolving Skill Libraries](https://mlsanity.com/paper/2607.10113/) (2026)
- [Towards a Systems Foundation for Agentic Skills: Architecture, Lifecycle, and Security](https://mlsanity.com/paper/2608.29596/) (2026)

## Abstract (arXiv)

Autonomous language-model agents increasingly rely on installable skills and tools to complete user tasks. Static skill auditing can expose capability surface before deployment, but it cannot determine whether a particular invocation is unsafe under the current user request and runtime context. We therefore study skill invocation auditing as a continuous-risk estimation problem: given a user request, candidate skill, and runtime context, predict a score that supports ranking and triage before a hard intervention is applied. We introduce STARS, which combines a static capability prior, a request-conditioned invocation risk model, and a calibrated risk-fusion policy. To evaluate this setting, we construct SIA-Bench, a benchmark of 3,000 invocation records with group-safe splits, lineage metadata, runtime context, canonical action labels, and derived continuous-risk targets. On a held-out split of indirect prompt injection attacks, calibrated fusion reaches 0.439 high-risk AUPRC, improving over 0.405 for the contextual scorer and 0.380 for the strongest static baseline, while the contextual scorer remains better calibrated with 0.289 expected calibration error. On the locked in-distribution test split, gains are smaller and static priors remain useful. The resulting claim is therefore narrower: request-conditioned auditing is most valuable as an invocation-time risk-scoring and triage layer rather than as a replacement for static screening. Code is available at https://github.com/123zgj123/STARS.