gcp-cloud-functions
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill demonstrates processing untrusted input from HTTP requests and cloud events without sanitization, creating a surface for indirect prompt injection.
- Ingestion points: The skill reads input from
request.args.get("name")andcloud_event.datainmain.py, as well asreq.query.nameinindex.js. - Boundary markers: No explicit delimiters or instructions are used to distinguish untrusted data from system prompts.
- Capability inventory: The functions have capabilities to write to Firestore databases (
db.collection("orders").document(...).set(...)) and call the Google Cloud Vision API for image analysis. - Sanitization: Input values are interpolated directly into response strings and used in database document paths without validation or escaping.
Audit Metadata