skills/rabbit-ivan/ivan-skills/pptx/Gen Agent Trust Hub

pptx

Pass

Audited by Gen Agent Trust Hub on Mar 8, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill relies on several third-party libraries for its core functionality, including 'markitdown' and 'defusedxml' for Python, and 'pptxgenjs', 'playwright', and 'sharp' for Node.js. These are standard, well-maintained packages sourced from official registries.\n- [COMMAND_EXECUTION]: To support file conversions and comparisons, the skill executes system utilities like 'soffice', 'pdftoppm', and 'git'. These subprocess calls are constructed using safe patterns that pass arguments as lists, effectively neutralizing shell injection risks.\n- [PROMPT_INJECTION]: Processing external PowerPoint files involves an inherent attack surface for indirect prompt injection. The skill manages this risk through a documented evidence chain: (1) Ingestion points: PPTX files are parsed in 'ooxml/scripts/unpack.py' and 'scripts/inventory.py'. (2) Boundary markers: Not used for structural XML/ZIP parsing. (3) Capability inventory: The skill can execute subprocesses and launch a headless browser via 'playwright'. (4) Sanitization: Proactively applied using 'defusedxml' for secure XML parsing.\n- [SAFE]: The implementation demonstrates a clear focus on security by using 'defusedxml' to protect against XXE attacks and employing sandboxed rendering via Playwright for agent-generated HTML content. No malicious patterns or data exfiltration behaviors were found.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 8, 2026, 11:01 AM