When Scanners Lie: Evaluator Instability in LLM Red-Teaming

Automated LLM vulnerability scanners are increasingly used to assess security risks by measuring different attack type success rates ASR. Yet the validity of these measurements hinges on an often-overlooked component: the evaluator who determines whether an attack has succeeded. In this study, we...

Prototype Pollution

expr-eval and expr-eval-fork is vulnerable to Prototype Pollution. The vulnerability is due to improper handling of JavaScript prototype-based inheritance in the eval interface, which allows an attacker with access to manipulate object prototypes and potentially achieve arbitrary code execution...

Arbitrary Code Injection

Overview Affected versions of this package are vulnerable to Arbitrary Code Injection through the SCRIPTSAFEPREEXEC definition in RangerRequestScriptEvaluator. An attacker can execute arbitrary OS commands by invoking scripts that rebuild the script context/engine e.g., via loadWithNewGlobal and...

CVE-2026-27961

Agenta is an open-source LLMOps platform. A Server-Side Template Injection SSTI vulnerability exists in versions prior to 0.86.8 in Agenta's API server evaluator template rendering. Although the vulnerable code lives in the SDK package, it is executed server-side within the API process when runni...

PYSEC-2026-7

Agenta is an open-source LLMOps platform. A Server-Side Template Injection SSTI vulnerability exists in versions prior to 0.86.8 in Agenta's API server evaluator template rendering. Although the vulnerable code lives in the SDK package, it is executed server-side within the API process when runni...

CVE-2026-27961

Agenta is an open-source LLMOps platform. A Server-Side Template Injection SSTI vulnerability exists in versions prior to 0.86.8 in Agenta's API server evaluator template rendering. Although the vulnerable code lives in the SDK package, it is executed server-side within the API process when runni...

PYSEC-2026-7

Agenta is an open-source LLMOps platform. A Server-Side Template Injection SSTI vulnerability exists in versions prior to 0.86.8 in Agenta's API server evaluator template rendering. Although the vulnerable code lives in the SDK package, it is executed server-side within the API process when runni...

EUVD-2026-8814

Agenta is an open-source LLMOps platform. In Agenta-API prior to version 0.48.1, a Python sandbox escape vulnerability existed in Agenta's custom code evaluator. Agenta used RestrictedPython as a sandboxing mechanism for user-supplied evaluator code, but incorrectly whitelisted the numpy package ...

CVE-2025-33246

NVIDIA NeMo Framework for all platforms contains a vulnerability in the ASR Evaluator utility, where a user could cause a command injection by supplying crafted input to a configuration parameter. A successful exploit of this vulnerability might lead to code execution, escalation of privileges,...

CVE-2025-33246

NVIDIA NeMo Framework for all platforms contains a vulnerability in the ASR Evaluator utility, where a user could cause a command injection by supplying crafted input to a configuration parameter. A successful exploit of this vulnerability might lead to code execution, escalation of privileges,...

CVE-2025-33246

NVIDIA NeMo Framework for all platforms contains a vulnerability in the ASR Evaluator utility, where a user could cause a command injection by supplying crafted input to a configuration parameter. A successful exploit of this vulnerability might lead to code execution, escalation of privileges,...

CVE-2025-33246

NVIDIA NeMo Framework for all platforms contains a vulnerability in the ASR Evaluator utility, where a user could cause a command injection by supplying crafted input to a configuration parameter. A successful exploit of this vulnerability might lead to code execution, escalation of privileges,...

CVE-2025-33246

NVIDIA NeMo Framework for all platforms contains a vulnerability in the ASR Evaluator utility, where a user could cause a command injection by supplying crafted input to a configuration parameter. A successful exploit of this vulnerability might lead to code execution, escalation of privileges,...

CVE-2025-33246

CVE-2025-33246 affects the NVIDIA NeMo Framework across platforms, with the vulnerability residing in the ASR Evaluator utility. A crafted input in a configuration parameter can trigger a command injection, enabling possible code execution, privilege escalation, data tampering, or information dis...

PT-2026-20405

Name of the Vulnerable Software and Affected Versions NVIDIA NeMo Framework affected versions not specified Description The NVIDIA NeMo Framework contains a flaw within the ASR Evaluator utility. A user can inject commands by providing specially crafted input to a configuration parameter...

NVIDIA Nemo Framework 命令注入漏洞

NVIDIA Nemo Framework is a framework developed by NVIDIA Corporation in the United States for building and deploying generative AI models. The NVIDIA NeMo Framework has a command injection vulnerability. This vulnerability stems from the ASR Evaluator tool, where users can exploit the framework b...

CVE-1999-0455

The Expression Evaluator sample application in ColdFusion allows remote attackers to read or delete files on the server via exprcalc.cfm, which does not restrict access to the server properly...

BIT-KIBANA-2025-68387 Kibana Improper Neutralization of Input During Web Page Generation ('Cross-site Scripting')

Improper neutralization of input during web page generation 'Cross-site Scripting' CWE-79 allows an unauthenticated user to embed a malicious script in content that will be served to web browsers causing cross-site scripting XSS CAPEC-63 via a vulnerability a function handler in the Vega AST...

BIT-ELK-2025-68387 Kibana Improper Neutralization of Input During Web Page Generation ('Cross-site Scripting')

Improper neutralization of input during web page generation 'Cross-site Scripting' CWE-79 allows an unauthenticated user to embed a malicious script in content that will be served to web browsers causing cross-site scripting XSS CAPEC-63 via a vulnerability a function handler in the Vega AST...

CVE-2025-68387

A flaw was found in Kibana. An unauthenticated user can embed a malicious script in web page content through improper input neutralization during web page generation. This cross-site scripting XSS vulnerability, specifically in a function handler within the Vega AST evaluator, allows for the...