MemSecBench: Tracking Agent Memory Poisoning from Persistence to Consequence and Repair

Memory systems allow agents to retain and reuse information from past interactions, but they can also let malicious content persist. A malicious instruction crafted by an attacker may be stored in long-term memory, recalled much later, and quietly shape a real action. Recent benchmarks increasing...

FlowGuard: From Signals to Evidence for MCP Security Detection

The Model Context Protocol MCP enables LLM agents to interact with external tools through metadata exchange, tool invocation, and response consumption. Existing MCP security scanners primarily reason about suspicious semantic signals rather than real execution behaviors, which can lead to...

Using LLMs to Adjudicate Static-Analysis Alerts with Error Reduction Techniques

Static analysis is widely used for finding security weaknesses in source code before deployment, but it often produces far more alerts than analysts can review. We study how well large language models LLMs can adjudicate classify as a real bug or a false alarm static-analysis alerts. We use two...

Malicious code in @checkrhq/adjudication-api-client (npm)

--- -= Per source details. Do not edit below this line.=- Source: amazon-inspector 8c52d9987ace09f0c48d8b0661bd1fcd3cf4e7e701b5e515810c7f32d99086cf package.json declares a preinstall lifecycle script that runs curl to POST installer reconnaissance data — hostname, current user whoami, working...

MAL-2026-6789 Malicious code in @checkrhq/adjudication-api-client (npm)

--- -= Per source details. Do not edit below this line.=- Source: amazon-inspector 8c52d9987ace09f0c48d8b0661bd1fcd3cf4e7e701b5e515810c7f32d99086cf package.json declares a preinstall lifecycle script that runs curl to POST installer reconnaissance data — hostname, current user whoami, working...

Structured Security Auditing and Robustness Enhancement for Untrusted Agent Skills

Agent Skills package SKILL.md files, scripts, reference documents, and repository context into reusable capability units, turning pre-load auditing from single-prompt filtering into cross-file security review. Existing guardrails often flag risk but recover malicious intent inconsistently under...

Second Interdisciplinary Workshop on Reimagining Democracy

Last month, I convened the Second Interdisciplinary Workshop on Reimagining Democracy IWORD 2023 at the Harvard Kennedy School Ash Center. As with IWORD 2022, the goal was to bring together a diverse set of thinkers and practitioners to talk about how democracy might be reimagined for the...

Hacking Trespass Law

This article talks about public land in the US that is completely surrounded by private land, which in some cases makes it inaccessible to the public. But theres a hack: Some hunters have long believed, however, that the publicly owned parcels on Elk Mountain can be legally reached using a practi...

Smart Contract Bug Results in $31 Million Loss

A hacker stole $31 million from the blockchain company MonoX Finance , by exploiting a bug in software the service uses to draft smart contracts. Specifically, the hack used the same token as both the tokenIn and tokenOut, which are methods for exchanging the value of one token for another. MonoX...

JudgeOnline open source adjudication system suffers from xss vulnerability

The Online Judge system OJ for short is an online system for judging questions. An xss vulnerability exists in the JudgeOnline open source judging system. An attacker can exploit this vulnerability to obtain an administrator cookie...

deniedboardingadjudication.info XSS vulnerability

Open Bug Bounty ID: OBB-702632 Description| Value ---|--- Affected Website:| deniedboardingadjudication.info Open Bug Bounty Program:| Create your bounty program now. It's open and free. Vulnerable Application:| hidden until disclosure Vulnerability Type:| XSS Cross Site Scripting / CWE-79 CVSSv3...