real-estate-property-matching-engine
Pass
Audited by Gen Agent Trust Hub on Nov 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external data from buyer profiles and property portals, creating a potential surface for indirect prompt injection.
- Ingestion points: Step 1 (Profile Intake) processes buyer requirements and inquiry patterns; Step 2 (Market Scan) involves searching external listing sources and pulling transaction data (SKILL.md).
- Boundary markers: The instructions do not define specific delimiters or instructions to ignore instructions embedded within the untrusted buyer or market data.
- Capability inventory: The skill uses a shell script (scripts/run.sh) to execute a platform-level runner and implies network operations to fetch market data from portals.
- Sanitization: There are no documented sanitization, validation, or filtering steps for the data ingested from external real estate listings or buyer inquiries.
Audit Metadata