gcp-cloud-run
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill documents templates that ingest untrusted data from external sources, creating a potential attack surface for indirect prompt injection or data manipulation.
- Ingestion points: Untrusted data enters the context via HTTP request parameters (
req.params.id), query strings (req.query.name), request bodies (req.body.name), and Google Cloud Pub/Sub messages (decoded from base64 strings inSKILL.md). - Boundary markers: The provided code snippets do not include explicit delimiters or instructions to ignore embedded commands within the processed data.
- Capability inventory: The skill includes capabilities for executing database queries via
@google-cloud/bigquery, performing logging, and serving web content. - Sanitization: The examples demonstrate basic processing such as
JSON.parsebut lack input validation or escaping mechanisms to sanitize external content before it is used in logic or database queries.
Audit Metadata