Issues

[UPDATE] agentic-top10/Agentic_NISTAIRMF.md - 5 NIST AI RMF subcategory IDs used with the wrong meaning (21 rows)

GenAI-Security-Project/crosswalk#217 · opened

on Oct 7, 2026

1

[new-cve] CVE-2026-51871 — Devika v1.0 is vulnerable to Code Injection in the Runner.execute function in sr

GenAI-Security-Project/crosswalk#212 · opened

on Oct 5, 2026

[new-research] arXiv:2610.02302 — Intent-Hiding Jailbreaks: An Information-Theoretic Framework for Compositional A

GenAI-Security-Project/crosswalk#210 · opened

on Oct 5, 2026

[new-research] arXiv:2610.01871 — Walking the Embedding Space: Datastore Extraction from Multimodal RAG

GenAI-Security-Project/crosswalk#211 · opened

on Oct 5, 2026

[new-research] arXiv:2610.02349 — MIRROR: Multipath Quorum Integrity for LLM Multi-Agent Communication

GenAI-Security-Project/crosswalk#209 · opened

on Oct 5, 2026

[new-research] arXiv:2610.02373 — Hop-Decayed Influence: New Vulnerabilities of Structural Auxiliary Indexing in G

GenAI-Security-Project/crosswalk#208 · opened

on Oct 5, 2026

[new-research] arXiv:2610.02418 — Mitigating Private Data Leakage in LLMs with Whiteout

GenAI-Security-Project/crosswalk#207 · opened

on Oct 5, 2026

[new-research] arXiv:2610.02456 — SideKernel: A Usable microVM Sandbox for AI Coding Agents on macOS

GenAI-Security-Project/crosswalk#205 · opened

on Oct 5, 2026

[new-research] arXiv:2610.02432 — Evaluating and Improving the Robustness of Large Language Models to Input Sequen

GenAI-Security-Project/crosswalk#206 · opened

on Oct 5, 2026

[new-research] arXiv:2610.02544 — CITADEL: CWE-Guided Insertion of Hardware Trojans via Analysis of DFG-Enabled LL

GenAI-Security-Project/crosswalk#204 · opened

on Oct 5, 2026

[new-research] arXiv:2610.02569 — Pincer: Resource Authorization for Agents using a Digital Twin

GenAI-Security-Project/crosswalk#203 · opened

on Oct 5, 2026

[new-research] arXiv:2610.02861 — Containing the Autonomous Operator: A Defense-in-Depth Framework and Reference A

GenAI-Security-Project/crosswalk#202 · opened

on Oct 5, 2026

[new-research] arXiv:2610.02910 — Frequency Is Not Sensitivity Identifying Safety-Sensitive Experts in Sparse MoE

GenAI-Security-Project/crosswalk#200 · opened

on Oct 5, 2026

[new-research] arXiv:2610.02869 — AgentTrap: Stateful Feedback Deception against Autonomous Penetration Testing Ag

GenAI-Security-Project/crosswalk#201 · opened

on Oct 5, 2026

[new-research] arXiv:2610.03089 — Securing Computer-Use Agents Against Branch Steering Attacks

GenAI-Security-Project/crosswalk#198 · opened

on Oct 5, 2026

[new-research] arXiv:2610.03014 — Beyond Predefined Sinks: Security-Aware Dependency Analysis for LLM Agents

GenAI-Security-Project/crosswalk#199 · opened

on Oct 5, 2026

[new-research] arXiv:2610.03153 — EvoRiskBench: An Evolving Benchmark for Runtime Security Risks in Workspace Agen

GenAI-Security-Project/crosswalk#196 · opened

on Oct 5, 2026

[new-research] arXiv:2610.03124 — The Fragility of Trigger-Tag Mechanisms for Misuse Detection in Open-Weight LLMs

GenAI-Security-Project/crosswalk#197 · opened

on Oct 5, 2026

[new-research] arXiv:2610.03434 — Persona Guardrail: A Production-Grade Defense Framework for Agentic Systems

GenAI-Security-Project/crosswalk#194 · opened

on Oct 5, 2026

[new-research] arXiv:2610.03319 — Defense-in-Depth at the Perception-Reasoning Interface of LLM-Centric Agentic UA

GenAI-Security-Project/crosswalk#195 · opened

on Oct 5, 2026

[new-research] arXiv:2610.03448 — Passing the Test You Trained On: Re-evaluating Prompt-Injection Detectors for LL

GenAI-Security-Project/crosswalk#193 · opened

on Oct 5, 2026

[new-research] arXiv:2610.03585 — Threat-Preserving Representation Sensitivity in Agent-Security Benchmarks

GenAI-Security-Project/crosswalk#192 · opened

on Oct 5, 2026

Data model for verification methods (follow-up to #101 / #186)

GenAI-Security-Project/crosswalk#191 · opened

on Oct 4, 2026

3

SME review: the three DRAFT entry mappings in evals/EXTERNAL_BENCHMARKS.md

GenAI-Security-Project/crosswalk#120 · opened

on Sep 18, 2026

1

[FRAMEWORK] NIST AI 600-1 (Generative AI Profile) — first addition on the ratified roadmap

GenAI-Security-Project/crosswalk#119 · opened

on Sep 18, 2026

2

Navigation Menu

Repository navigation

Issues

[UPDATE] agentic-top10/Agentic_NISTAIRMF.md - 5 NIST AI RMF subcategory IDs used with the wrong meaning (21 rows)

[new-cve] CVE-2026-51871 — Devika v1.0 is vulnerable to Code Injection in the Runner.execute function in sr

[new-research] arXiv:2610.02302 — Intent-Hiding Jailbreaks: An Information-Theoretic Framework for Compositional A

[new-research] arXiv:2610.01871 — Walking the Embedding Space: Datastore Extraction from Multimodal RAG

[new-research] arXiv:2610.02349 — MIRROR: Multipath Quorum Integrity for LLM Multi-Agent Communication

[new-research] arXiv:2610.02373 — Hop-Decayed Influence: New Vulnerabilities of Structural Auxiliary Indexing in G

[new-research] arXiv:2610.02418 — Mitigating Private Data Leakage in LLMs with Whiteout

[new-research] arXiv:2610.02456 — SideKernel: A Usable microVM Sandbox for AI Coding Agents on macOS

[new-research] arXiv:2610.02432 — Evaluating and Improving the Robustness of Large Language Models to Input Sequen

[new-research] arXiv:2610.02544 — CITADEL: CWE-Guided Insertion of Hardware Trojans via Analysis of DFG-Enabled LL

[new-research] arXiv:2610.02569 — Pincer: Resource Authorization for Agents using a Digital Twin

[new-research] arXiv:2610.02861 — Containing the Autonomous Operator: A Defense-in-Depth Framework and Reference A

[new-research] arXiv:2610.02910 — Frequency Is Not Sensitivity Identifying Safety-Sensitive Experts in Sparse MoE

[new-research] arXiv:2610.02869 — AgentTrap: Stateful Feedback Deception against Autonomous Penetration Testing Ag

[new-research] arXiv:2610.03089 — Securing Computer-Use Agents Against Branch Steering Attacks

[new-research] arXiv:2610.03014 — Beyond Predefined Sinks: Security-Aware Dependency Analysis for LLM Agents

[new-research] arXiv:2610.03153 — EvoRiskBench: An Evolving Benchmark for Runtime Security Risks in Workspace Agen

[new-research] arXiv:2610.03124 — The Fragility of Trigger-Tag Mechanisms for Misuse Detection in Open-Weight LLMs

[new-research] arXiv:2610.03434 — Persona Guardrail: A Production-Grade Defense Framework for Agentic Systems

[new-research] arXiv:2610.03319 — Defense-in-Depth at the Perception-Reasoning Interface of LLM-Centric Agentic UA

[new-research] arXiv:2610.03448 — Passing the Test You Trained On: Re-evaluating Prompt-Injection Detectors for LL

[new-research] arXiv:2610.03585 — Threat-Preserving Representation Sensitivity in Agent-Security Benchmarks

Data model for verification methods (follow-up to #101 / #186)

SME review: the three DRAFT entry mappings in evals/EXTERNAL_BENCHMARKS.md

[FRAMEWORK] NIST AI 600-1 (Generative AI Profile) — first addition on the ratified roadmap