wandb-experiment-logger
Pass
Audited by Gen Agent Trust Hub on Sep 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill presents an indirect prompt injection surface due to its auto-activation mechanism and tool access.
- Ingestion points: The skill activates automatically when processing user-supplied keywords such as "wandb", "wandb experiment logger", or "wandb logger" (SKILL.md).
- Boundary markers: There are no explicit instructions or delimiters defined to prevent the agent from following malicious instructions that might be embedded in the training data, code snippets, or configuration files being processed.
- Capability inventory: The skill utilizes
Write,Edit,Bash(python:*), andBash(pip:*)tools (SKILL.md), which allow for modification of the local environment. - Sanitization: No sanitization or validation logic is defined to filter instructions from untrusted external data sources before they reach the agent context.
Audit Metadata