k9-ownership-byod-setup
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-defined organizational context and tag values intended for consumption by an AI inference engine. The implementation includes specific security controls to mitigate risks.
- Ingestion points: Customer-defined values in
assets/example.csvand the structured schema inreferences/schema.md. - Boundary markers: The skill employs a strict CSV schema to isolate data fields.
- Capability inventory: The agent uses
ReadandBashtools for file access and utility execution. - Sanitization:
references/schema.mddefines robust validation rules, explicitly forbidding characters like angle brackets, curly braces, and pipes to prevent prompt injection.
Audit Metadata