Advisory Catalog

Open Neural Network Exchange (ONNX) Path Traversal Vulnerability

Aim vulnerable to Cross-Site Request Forgery

Aim External Control of File Name or Path vulnerability

llama-index-packs-finchat SQL Injection vulnerability

LlamaIndex Retrievers Integration: DuckDBRetriever SQL Injection

InvokeAI Deserialization of Untrusted Data vulnerability

InvokeAI Arbitrary File Deletion vulnerability

vLLM Deserialization of Untrusted Data vulnerability

DB-GPT vulnerable to Arbitrary File Upload with Path Traversal

DB-GPT Arbitrary File Write vulnerability

DB-GPT is vulnerable to SQL Injection attacks from unauthenticated users

DB-GPT Absolute Path Traversal in knowledge/{space_name}/document/upload

DB-GPT Absolute Path Traversal vulnerability

H2O Deserialization of Untrusted Data Vulnerability

Horovod Vulnerable to Command Injection

vLLM Allows Remote Code Execution via Mooncake Integration

Qiskit allows arbitrary code execution decoding QPY format versions < 13

PlotAI eval vulnerability

Spacy-LLM Server-Side Template Injection (SSTI) vulnerability

LTI JupyterHub Authenticator does not properly validate JWT Signature

PandasAI interactive prompt function Remote Code Execution (RCE)

Django-Unicorn Class Pollution Vulnerability, Leading to XSS, DoS and Authentication Bypass

Sentry's improper authentication on SAML SSO process allows user impersonation

Gradio Blocked Path ACL Bypass Vulnerability

Rasa Allows Remote Code Execution via Remote Model Loading

unstructured XML External Entity (XXE)

Django SQL injection in HasKey(lhs, rhs) on Oracle

Deserialization of untrusted data in IPC and Parquet readers in the Apache Arrow R package versions 4.0.0 through 16.1.0 allows arbitrary code execution. An application is vulnerable if it reads Arrow IPC, Feather or Parquet data from untrusted sources (for example, user-supplied input files). Thi

libre-chat Path Traversal vulnerability

LLama Factory Remote OS Command Injection Vulnerability

cobbler allows anyone to connect to cobbler XML-RPC server with known password and make changes

codechecker vulnerable to authentication bypass when using specifically crafted URLs

codechecker authentication method confusion vulnerability allows logging in as the built-in root user from an external service

Langflow vulnerable to remote code execution

AgentScope uses `eval`

langflow has vulnerability in PythonCodeTool component

In PyTorch <=2.4.1, the RemoteModule has Deserialization RCE. NOTE: this is disputed by multiple parties because this is intended behavior in PyTorch distributed computing.

Waitress has request processing race condition in HTTP pipelining with invalid first request

Langchain SQL Injection vulnerability

Lollms vulnerable to Cross-site Scripting

LangChain Experimental Eval Injection vulnerability

sqlitedict insecure deserialization vulnerability

AutoGPT bypass of the shell commands denylist settings

D-Tale vulnerable to Remote Code Execution through the Query input on Chart Builder

pyload-ng vulnerable to RCE with js2py sandbox escape

H2O.ai H2O vulnerable to deserialization attacks via a JDBC Connection URL

MindsDB Vulnerable to Bypass of SSRF Protection with DNS Rebinding

LlamaIndex includes an exec call for `import {cls_name}`

Mobile Security Framework (MobSF) has a Zip Slip Vulnerability in .a Static Library Files

Django SQL injection vulnerability