structured-logging
Pass
Audited by Gen Agent Trust Hub on Sep 28, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill defines patterns for logging metadata and request headers originating from external sources. This creates a vulnerability surface where untrusted data could be interpolated into logs or propagated to other services.\n
- Ingestion points: Extraction of
x-request-idfrom request headers and acceptance ofRecord<string, unknown>as metadata inContextLoggercalls (SKILL.md).\n - Boundary markers: The skill notes that the error middleware normalizes values and hides internal details from clients in production.\n
- Capability inventory: The resulting code performs file/stdout writes (via the logger transport) and includes network operations through header propagation in repository calls.\n
- Sanitization: No explicit instructions are provided for sanitizing or escaping the content of metadata or event messages before they are logged.
Audit Metadata