conversion-copywriter
Pass
Audited by Gen Agent Trust Hub on Aug 12, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from customer reviews and external website URLs to harvest facts. Ingestion occurs in SKILL.md (§0 and §1) during the intake phase. Boundary markers are included in the instructions, directing the agent to use external content only for fact mining and to avoid mirroring the original phrasing. The skill utilizes Read, Write, WebFetch, and Bash tools. All data rendered in the final report is sanitized using HTML escaping in scripts/render_deck.py.
- [COMMAND_EXECUTION]: The skill uses a local script, scripts/render_deck.py, to generate an HTML copy deck from JSON data. The script relies solely on the Python standard library and performs legitimate local file system operations.
- [EXTERNAL_DOWNLOADS]: The skill uses WebFetch to retrieve content from user-provided URLs for the purpose of fact extraction, which is consistent with its primary function.
Audit Metadata