frevana-gen-report
Pass
Audited by Gen Agent Trust Hub on Jul 15, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill facilitates an indirect prompt injection surface by ingesting untrusted data (via
--contentor--content-file) and passing it to an external API to generate HTML. While the skill correctly uses JSON escaping and avoids direct evaluation, the resulting HTML could contain malicious instructions or scripts that might influence downstream agents or user browsers. - Ingestion points:
scripts/generate_report.shreads user content from command-line arguments or local files. - Boundary markers: None detected in the script or prompt instructions to warn the agent about ignoring instructions embedded within the processed content.
- Capability inventory: The skill uses
curlfor network requests andpython3for JSON parsing and file writing. - Sanitization: The script performs standard JSON escaping for the outbound request, but it does not sanitize or filter the HTML returned by the API before outputting it or saving it to a file.
Audit Metadata