skills/mukul975/anthropic-cybersecurity-skills/implementing-application-whitelisting-with-applocker/Gen Agent Trust Hub
implementing-application-whitelisting-with-applocker
Pass
Audited by Gen Agent Trust Hub on Sep 22, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The analysis scripts in
scripts/agent.pyandscripts/process.pyingest untrusted external data in the form of AppLocker policy XML files and event log exports (CSV/JSON). - Ingestion points:
scripts/agent.pyprocesses XML policy files and JSON-formatted event logs;scripts/process.pyprocesses CSV event logs. - Boundary markers: The scripts do not implement specific delimiters or safety instructions to distinguish between data and potential commands within the files.
- Capability inventory: Analysis is limited to file system read/write operations for reporting; no network operations, command execution, or dynamic code evaluation were identified.
- Sanitization: The scripts rely on standard library parsers (
xml.etree.ElementTree,json,csv) without additional sanitization of the input content. - [EXTERNAL_DOWNLOADS]: The documentation and standards references include links to external resources for security research and policy development, such as Microsoft's Learn platform, NIST, the Australian Cyber Security Centre (ACSC), and the LOLBAS project.
Audit Metadata