skills/mukul975/anthropic-cybersecurity-skills/performing-privilege-escalation-assessment/Gen Agent Trust Hub
performing-privilege-escalation-assessment
Warn
Audited by Gen Agent Trust Hub on Mar 20, 2026
Risk Level: MEDIUMCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSDATA_EXFILTRATIONREMOTE_CODE_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The
scripts/agent.pyscript utilizessubprocess.runwithshell=Trueto perform system enumeration tasks. This pattern allows for the execution of arbitrary shell commands and is vulnerable to manipulation if the system output being processed (such as filenames or script paths in cron jobs) contains malicious shell metacharacters. - [EXTERNAL_DOWNLOADS]: The skill provides instructions and references for downloading and running external scripts and binaries from third-party GitHub repositories (e.g.,
linPEAS,winPEAS,PrintSpoofer), which are executed to automate the discovery of escalation paths. - [DATA_EXFILTRATION]: The assessment workflow includes instructions for the agent to access and extract highly sensitive system data, including password hashes from
/etc/shadow, credentials from.envfiles, and cloud instance metadata containing IAM roles from the link-local address169.254.169.254. - [REMOTE_CODE_EXECUTION]: The documentation provides detailed steps for generating malicious MSI packages via
msfvenomand executing them on target systems to achieve reverse shells and SYSTEM-level privileges.
Audit Metadata