HIGH
GHSASUPPLY-CHAINLLM01: Prompt Injection
MCP-for-Stata: Stata Command Injection via Unsanitized `package` in `ado_package_install`
Unsanitized MCP input can trigger Stata shell escapes and run OS commands as the server user.
7 HIGH · 2 MED · 1 LOW · 1 INFO · THREAT ORANGE · 11 items · Generated in 255s
Unsanitized MCP input can trigger Stata shell escapes and run OS commands as the server user.
Unvalidated MongoDB operators let a low-privileged caller alter queries and expose other tenants' data.
Default SSRF allows internal or metadata fetches through the server, exposing sensitive services and data.
Unauthenticated SSRF can reach internal or cloud metadata services and expose response data from vulnerable MLflow servers.
A flawed auth check can let one user create runs against another user's private assistant and expose its metadata.
An attacker or prompt-injected client can read or write local files the server user can access, including secrets and configs.
A prompt-injected or malicious MCP client can read or write local files the server user can access, exposing secrets or altering data.
Unauthenticated path traversal can expose session log files from outside the intended directory.
This is a recap headline, not a specific flaw. It signals broad AI-related attack activity worth tracking and triaging.
Signals rising vulnerability discovery pressure and slower remediation, which increases backlog and exposure risk.
This is general analysis about AI-era app security, not a specific exploit, breach, or vulnerable product.