From 11cfec1d046a95a47d0d3452d4ee2b650d5558d2 Mon Sep 17 00:00:00 2001 From: vgudur-dev Date: Mon, 26 Oct 2026 11:08:14 -0700 Subject: [PATCH 1/4] Add OWASP Agent Memory Guard and AgentThreatBench to ASI06 tools Signed-off-by: vgudur-dev --- data/tools-supplement.json | 10 ++++++++++ 1 file changed, 10 insertions(+) diff --git a/data/tools-supplement.json b/data/tools-supplement.json index 85c0a72..5d16b6a 100644 --- a/data/tools-supplement.json +++ b/data/tools-supplement.json @@ -94,6 +94,16 @@ "name": "LAAF v2.0", "url": "https://github.com/qorvexconsulting1/laaf-V2.0", "type": "open-source" + }, + { + "name": "OWASP Agent Memory Guard", + "url": "https://github.com/OWASP/www-project-agent-memory-guard", + "type": "open-source" + }, + { + "name": "AgentThreatBench (Inspect Evals)", + "url": "https://github.com/UKGovernmentBEIS/inspect_evals/tree/main/src/inspect_evals/agent_threat_bench", + "type": "open-source" } ], "ASI07": [ From e5ecf7e0b4a988784601f1e2b1c0e592ad7d8071 Mon Sep 17 00:00:00 2001 From: vgudur-dev Date: Mon, 26 Oct 2026 11:08:27 -0700 Subject: [PATCH 2/4] Add OWASP Agent Memory Guard and AgentThreatBench to ASI06 tools Signed-off-by: vgudur-dev --- data/entries/ASI06.json | 10 ++++++++++ 1 file changed, 10 insertions(+) diff --git a/data/entries/ASI06.json b/data/entries/ASI06.json index 9d577d8..630371b 100644 --- a/data/entries/ASI06.json +++ b/data/entries/ASI06.json @@ -1086,6 +1086,16 @@ "name": "LAAF v2.0", "url": "https://github.com/qorvexconsulting1/laaf-V2.0", "type": "open-source" + }, + { + "name": "OWASP Agent Memory Guard", + "url": "https://github.com/OWASP/www-project-agent-memory-guard", + "type": "open-source" + }, + { + "name": "AgentThreatBench (Inspect Evals)", + "url": "https://github.com/UKGovernmentBEIS/inspect_evals/tree/main/src/inspect_evals/agent_threat_bench", + "type": "open-source" } ], "incidents": [ From 6606ca47cf2657ec0700ac2143fcccf8a3d5d7d8 Mon Sep 17 00:00:00 2001 From: vgudur-dev Date: Mon, 26 Oct 2026 11:09:47 -0700 Subject: [PATCH 3/4] Add OWASP Agent Memory Guard and AgentThreatBench to ASI06 tools Signed-off-by: vgudur-dev --- docs/data.js | 10 ++++++++++ 1 file changed, 10 insertions(+) diff --git a/docs/data.js b/docs/data.js index f5cf848..518c70c 100644 --- a/docs/data.js +++ b/docs/data.js @@ -20452,6 +20452,16 @@ window.CROSSWALK_DATA = [ "name": "LAAF v2.0", "url": "https://github.com/qorvexconsulting1/laaf-V2.0", "type": "open-source" + }, + { + "name": "OWASP Agent Memory Guard", + "url": "https://github.com/OWASP/www-project-agent-memory-guard", + "type": "open-source" + }, + { + "name": "AgentThreatBench (Inspect Evals)", + "url": "https://github.com/UKGovernmentBEIS/inspect_evals/tree/main/src/inspect_evals/agent_threat_bench", + "type": "open-source" } ], "incidents": [ From 40065dd0664574aa7fe9cd60a3c8b8aede5b32cb Mon Sep 17 00:00:00 2001 From: vgudur-dev Date: Wed, 28 Oct 2026 11:33:58 -0700 Subject: [PATCH 4/4] evals: add AgentThreatBench to EXTERNAL_BENCHMARKS.md Added AgentThreatBench to evaluate LLM agents' performance against adversarial content while maintaining utility and security. Signed-off-by: vgudur-dev --- evals/EXTERNAL_BENCHMARKS.md | 2 ++ 1 file changed, 2 insertions(+) diff --git a/evals/EXTERNAL_BENCHMARKS.md b/evals/EXTERNAL_BENCHMARKS.md index dba1f7d..8b10496 100644 --- a/evals/EXTERNAL_BENCHMARKS.md +++ b/evals/EXTERNAL_BENCHMARKS.md @@ -45,6 +45,7 @@ as a profile in this repository.** | **AgentLSD** | **Adversarial task contamination** of AI security agents — deceptive artifacts in the environment, including non-instructional evidence "such as fake results and decoy endpoints", rather than injected instructions alone | Paired clean and trap-augmented runs on CTF challenges with deterministic trap generation; 6 models on 11 web CTF challenges | LLM01 Prompt Injection · ASI01 Agent Goal Hijack | [arXiv:2609.19140](https://arxiv.org/abs/2609.19140) · release: [Golim/agent-lsd](https://github.com/Golim/agent-lsd) (MIT) | | **AgentXploit-Bench** | **End-to-end exploitability of AI-agent systems** in authorized white-box pre-deployment auditing: attacks must act through the task-defined attacker interface and be confirmed by an external verifier | 72 reproducible vulnerabilities across 12 open-source AI-agent systems and frameworks | ASI02 Tool Misuse and Exploitation · ASI05 Unexpected Code Execution · LLM01 Prompt Injection | [arXiv:2609.31318](https://arxiv.org/abs/2609.31318) · release: [lwd17/AgentXploit](https://github.com/lwd17/AgentXploit) (README states Apache-2.0; no licence file in the repository) | | **PrivDrift** | **User-secret leakage under topic drift**: whether secrets a user disclosed earlier in an active conversation remain recoverable after the dialogue moves on, under persuasion-based probing | 1,000 controlled multi-turn dialogues with seeded secrets, content-dense drift turns and standardized extraction probes; 3 LLMs with extended context windows | LLM02 Sensitive Information Disclosure · DSGAI11 Cross-Context Conversation Bleed | [arXiv:2609.30094](https://arxiv.org/abs/2609.30094) · not released: the paper states the authors "plan to release" the generation code, probes, scripts and a sanitized subset | +| **AgentThreatBench** | Whether **LLM agents complete useful work while resisting adversarial content** in tool outputs and memory stores, scored separately for utility and security | 3 Inspect AI tasks: memory poisoning (10 samples), email-triage goal hijack (6 samples), customer-support data exfiltration (8 samples); each includes clean baselines | ASI06 Memory and Context Poisoning · ASI01 Agent Goal Hijack | release: [UKGovernmentBEIS/inspect_evals, agent_threat_bench](https://github.com/UKGovernmentBEIS/inspect_evals/tree/main/src/inspect_evals/agent_threat_bench) (MIT). Contributed by the PR author, who also co-leads OWASP Agent Memory Guard. | ## Why these and not others @@ -56,3 +57,4 @@ Routing — incident, catalogue, or noted and closed — is defined once in [`.. |---|---| | 2026-09-18 | Created with LongPIBench, GenIaC-SecBench and TIER, from the watcher triage of issues #47, #55 and #70. | | 2026-09-30 | Added APort Vault, ClashBench, AgentLSD, AgentXploit-Bench and PrivDrift, from the watcher triage of issues #125, #137, #144, #160 and #171; added the release and licence note. | +| 2026-10-28 | Added AgentThreatBench, from PR #321. |