audit
Pass
Audited by Gen Agent Trust Hub on Apr 15, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill displays content from audit logs and decision traces, which constitutes a surface for indirect prompt injection.
- Ingestion points: Data is ingested from logs stored at
.bkit/audit/YYYY-MM-DD.jsonland.bkit/decisions/YYYY-MM-DD.jsonl(SKILL.md). - Boundary markers: The skill does not define explicit boundary markers or instructions to ignore embedded instructions when parsing and displaying log content.
- Capability inventory: The skill has access to
Bash,Grep,Read, andGlobtools (SKILL.md). - Sanitization: No sanitization, escaping, or validation of the log content is performed before presenting the data to the agent context.
Audit Metadata