audit-service-gaps
Pass
Audited by Gen Agent Trust Hub on Sep 20, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to ingest and analyze external content such as reference service source code and documentation. This creates an attack surface where malicious instructions embedded in the analyzed code could potentially influence the agent's audit findings.
- Ingestion points:
docs/NORTH_STAR.md, external reference service repositories, and third-party documentation. - Boundary markers: No explicit instruction for the agent to use delimiters or ignore instructions within external data is provided.
- Capability inventory: The skill involves reading files and writing Audit Decision Records (ADRs) to the local filesystem (
docs/decisions/). It does not specify tool-based execution of external code. - Sanitization: No explicit sanitization or validation of the external content is mentioned.
- [SAFE]: The skill uses well-known industry frameworks (Heuristic Evaluation, Jobs-to-be-Done, North Star Framework) to guide the agent through a logical auditing process.
- [SAFE]: The inclusion of Korean language prompts is legitimate and matches the stated purpose of supporting users who use specific terminology for gap analysis.
Audit Metadata