perimeter
Pass
Audited by Gen Agent Trust Hub on Nov 5, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests and audits untrusted source code from a target directory, which serves as a potential vector for malicious instructions embedded in the code being analyzed.
- Ingestion points: Phase 1 (Macro Anchor Grounding) and Phase 2 (Scale-Adaptive Boundary Audit) ingest target directories and subsystems for architectural scouting.
- Boundary markers: The skill uses a 'Load-Bearing Qualification Gate' to qualify findings and a 'Canonical Output Contract' to structure reports, but it does not specify explicit delimiters or 'ignore' instructions for the code ingestion phase.
- Capability inventory: The skill provides routing instructions to external tools such as
/changesetand/forgefor automated code modifications based on audit results. - Sanitization: There is no evidence of explicit sanitization or filtering of the source code content before it is processed by the agent's logic.
Audit Metadata