High AUC from linear probes on model activations for indirect prompt injection does not license an unqualified claim of malicious-content detection, per a Qwen2.5-VL-7B case study with text and visual controls.
SOK: A Taxonomy of Attack Vectors and Defense Strategies for Agentic Supply Chain Runtime
7 Pith papers cite this work. Polarity classification is still indexing.
7
Pith papers citing it
abstract
Agentic systems based on large language models (LLMs) operate not merely as text generators but as autonomous entities that dynamically retrieve information and invoke tools. This execution model shifts the attack surface from traditional build-time artifacts to inference-time dependencies, exposing agents to manipulation through untrusted data and probabilistic capability resolution. While prior work has examined model-level vulnerabilities, security risks arising from the complex, cyclic runtime behavior of agents remain fragmented. This paper systematizes existing research into a unified runtime framework. We categorize threats into data supply chain attacks (distinguishing between transient context injection and persistent memory poisoning) and tool supply chain attacks (spanning discovery, implementation, and invocation phases). Crucially, we identify the emergence of the Viral Agent Loop, where agents effectively become vectors for self-propagating generative worms that require no code vulnerabilities to spread. We argue for a transition to a Zero-Trust Runtime Architecture, where context is treated as untrusted control flow, and tool execution is bounded by cryptographic provenance rather than semantic likelihood.
citation-role summary
background 3
citation-polarity summary
years
2026 7roles
background 3polarities
background 3representative citing papers
Large-scale audit of SkillsMP agent skills finds 520 skills with 1,708 credential-leak issues, dominated by debug logging into the LLM context and hard-to-remediate forks.
About 18.2% of structurally flagged skill pairs represent genuine compositional safety risks in agent skill registries, with exploitation gated by host model behavior.
ChainCaps uses monotonic capability attenuation via intersection of sink-specific budgets in a transparent proxy to reduce attack success on composed tool-using agents from 25-68% to 0-4.8% while keeping 96-100% benign task completion.
Reflect-Guard fine-tunes Llama-Guard-3-8B with distilled self-reflections to raise F1 on WildGuardTest from 0.770 to 0.842 and cut JailbreakBench attack success from 10.3% to 1.8%.
A malicious BYOK relay can rewrite an LLM agent's execution-bearing response fields after safety alignment, achieving 73.5-99.1% attack success on agent benchmarks while bypassing model defenses.
A synthesis of 247 papers on LLM agent security identifies prompt injection and tool hijacking as dominant threats, notes weakly compositional defenses, and argues for trust boundaries and realistic evaluations.
citing papers explorer
-
When AUC 0.998 Is Not Enough: A Candidate Evaluation Protocol for Hidden-State Probes of Indirect Prompt Injection in Multimodal Computer-Use Agents
High AUC from linear probes on model activations for indirect prompt injection does not license an unqualified claim of malicious-content detection, per a Qwen2.5-VL-7B case study with text and visual controls.
-
How Your Credentials Are Leaked by LLM Agent Skills: An Empirical Study
Large-scale audit of SkillsMP agent skills finds 520 skills with 1,708 credential-leak issues, dominated by debug logging into the LLM context and hard-to-remediate forks.
-
When Safe Skills Collide: Measuring Compositional Risk in Agent Skill Ecosystems
About 18.2% of structurally flagged skill pairs represent genuine compositional safety risks in agent skill registries, with exploitation gated by host model behavior.
-
ChainCaps: Composition-Safe Tool-Using Agents via Monotonic Capability Attenuation
ChainCaps uses monotonic capability attenuation via intersection of sink-specific budgets in a transparent proxy to reduce attack success on composed tool-using agents from 25-68% to 0-4.8% while keeping 96-100% benign task completion.
-
Reflect-Guard: Enhancing LLM Safeguards against Adversarial Prompts via Logical Self-Reflection
Reflect-Guard fine-tunes Llama-Guard-3-8B with distilled self-reflections to raise F1 on WildGuardTest from 0.770 to 0.842 and cut JailbreakBench attack success from 10.3% to 1.8%.
-
Rewriting the Response Path: Silent Tampering and Provider-Signed Defense in BYOK LLM Agents
A malicious BYOK relay can rewrite an LLM agent's execution-bearing response fields after safety alignment, achieving 73.5-99.1% attack success on agent benchmarks while bypassing model defenses.
-
Toward Secure LLM Agents: Threat Surfaces, Attacks, Defenses, and Evaluation
A synthesis of 247 papers on LLM agent security identifies prompt injection and tool hijacking as dominant threats, notes weakly compositional defenses, and argues for trust boundaries and realistic evaluations.