ai-core/debug-logging
Pass
Audited by Gen Agent Trust Hub on Sep 27, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill documents a logging system that captures raw data from AI providers and model outputs, which is a potential surface for indirect injection if downstream log processors are vulnerable.
- Ingestion points: The
providerandoutputcategories log raw frames and result content directly from external sources into the agent context (SKILL.md). - Boundary markers: Log lines use emoji and category prefixes (e.g.,
[tanstack-ai:provider]), though no explicit sanitization for the content of the logs is enforced by the skill instructions. - Capability inventory: The skill demonstrates piping logs to standard libraries like
pinoorwinston, which possess file-system and network capabilities. - Sanitization: The documentation provides best-practice examples for implementing a guarded logger to handle serialization safely when processing metadata.
- [EXTERNAL_DOWNLOADS]: The skill refers to official vendor packages
@tanstack/aiand@tanstack/ai-openai, as well as the industry-standard logging librarypino. These are well-known and trusted resources within the development ecosystem.
Audit Metadata