The Agent Skills Directory

[COMMAND_EXECUTION]: The agent.py script executes PowerShell commands via subprocess.run to interact with the Windows operating system for mutex management.
Evidence: check_mutex_exists_windows and create_mutex_windows functions in scripts/agent.py.
[REMOTE_CODE_EXECUTION]: The PowerShell command strings are constructed using direct string interpolation of the mutex_name variable, leading to a potential command injection vulnerability.
Evidence: In scripts/agent.py, the variable ps_script is built using f'try {{ $m = [System.Threading.Mutex]::OpenExisting("{mutex_name}"); ...'.
Risk: If an attacker provides a crafted mutex name (e.g., through a malicious malware sample analysis request), they could execute arbitrary shell commands on the host system.
[EXTERNAL_DOWNLOADS]: The skill references several external resources for threat intelligence and malware analysis.
Evidence: References to VirusTotal, ANY.RUN, and the community-maintained albertzsigovits/malware-mutex repository on GitHub in SKILL.md and references/api-reference.md. These are well-known technology and research services.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection due to its processing of untrusted data (ransomware indicators) without sanitization.
Ingestion points: The mutex_name and domain parameters in scripts/agent.py which may be populated from external malware samples or analysis reports.
Boundary markers: Absent. Data is interpolated directly into command strings and logic.
Capability inventory: Shell command execution via subprocess.run and DNS resolution via socket.gethostbyname in scripts/agent.py.
Sanitization: Absent. The script performs no validation or escaping of the input variables before use.

implementing-ransomware-kill-switch-detection