skill-usability-test
Pass
Audited by Gen Agent Trust Hub on Oct 24, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill requires the primary agent to audit transcripts and interview responses produced by a subagent, creating a data ingestion pipeline that processes untrusted content.
- Ingestion points: Transcripts and interview results are ingested for auditing in the 'インタビュー' and '結果の検証' sections (SKILL.md).
- Boundary markers: The instructions do not define specific delimiters or instructions to ignore embedded commands within the audited transcripts, which could lead the primary agent to inadvertently follow instructions found in the subagent's logs.
- Capability inventory: The agent is directed to audit tool invocation sequences and inspect repository file states, requiring file system read access.
- Sanitization: The skill lacks procedures for sanitizing or filtering the subagent's output before the auditing phase.
Audit Metadata