hwpx
Pass
Audited by Gen Agent Trust Hub on Aug 30, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill orchestrates internal workflows by using
subprocess.runto call helper scripts likebuild_hwpx.py,fix_namespaces.py, andvalidate.py. Commands are executed as argument lists, which prevents shell-based injection attacks. The execution is limited to the skill's bundled utilities and standard system binaries. - [INDIRECT_PROMPT_INJECTION]: The skill parses untrusted document content (HTML, Markdown, HWP) which could theoretically contain malicious instructions intended to influence the agent.
- Ingestion points: Files are ingested through
scripts/html2hwpx.py,scripts/text_extract.py, andscripts/convert_hwp.py. - Boundary markers: Content is escaped using
xml_escapebefore being placed in XML structures, providing protection against structural injection. - Capability inventory: The skill performs file system operations and subprocess execution based on its defined workflows.
- Sanitization: The skill uses defined templates and structural parsing to limit the impact of potentially malicious content in source documents.
Audit Metadata