expert-interview-topics
Pass
Audited by Gen Agent Trust Hub on Sep 23, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill includes a pre-generated audit report file (
expert-interview-topics_audit_result_v2.json) that explicitly claims the skill is "SAFE" and "Production Ready". This is self-referential content that attempts to influence the security evaluation. - [INDIRECT_PROMPT_INJECTION]: The script
scripts/main.pyingests external data from user-provided files via the--fileargument and interpolates it directly into LLM prompt templates. - Ingestion points: The
text_contentvariable inscripts/main.pyis populated by reading a local file path provided by the user. - Boundary markers: The prompt templates use labels like "Reference Text: {text}" but lack explicit instructions for the agent to ignore any malicious commands embedded within that text.
- Capability inventory: The script's functionality is limited to formatting and printing prompts to the standard output. No network operations, file-system modifications, or subprocess executions were found in the script logic.
- Sanitization: There is no evidence of input validation, escaping, or filtering applied to the content read from the external file before it is included in the output prompts.
Audit Metadata