approvals-reset-on-upload-v4-triage
Pass
Audited by Gen Agent Trust Hub on Apr 9, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill defines an investigation workflow that ingests untrusted data (such as contract IDs, workflow IDs, and request IDs) and interpolates them into SQL templates and logging filters. This presents a surface for indirect prompt injection or potential SQL injection if the underlying execution environment does not provide automatic parameterization.
- Ingestion points: User-provided variables such as
{contract_id},{wsid},{dataset}, and{request_id}referenced inSKILL.md. - Boundary markers: None present; the skill does not instruct the agent to use delimiters or ignore instructions embedded within the identifiers.
- Capability inventory: The skill utilizes BigQuery SQL execution and GCP Cloud Logging searches across multiple regional datasets.
- Sanitization: No sanitization or validation steps are defined for the input variables before they are used in query construction.
Audit Metadata