Observed arrival · 2026-11-12
AI2Shell’s 13/10 Security Advisory
Credibility concern recorded. The source reference remains available for verification and correction.
An advisory-style page frames giving AI agents unrestricted shell access as a command-execution vulnerability, with a section labeled “Simulation shell.”
- For
- Readers tracking AI-agent security and shell access
- Worth noticing
- The page assigns the issue a CVSS 4.0 score of 13.0/10 and calls it actively exploited; those claims are not independently substantiated here.
Field notes
The technical section describes agents turning natural-language requests into shell commands and running them with the permissions they were granted. A visible reproduction excerpt sets the shell tool to auto-accept, then asks the agent to tidy a repository. The page also names detection and mitigation sections, but the supplied extract does not verify its claims of active exploitation or official vulnerability listings.
Observed signals
Read the marks
Editorial observations of this landing page, not a rating.
One card from the complete issue