evidence
Pass
Audited by Gen Agent Trust Hub on Oct 10, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill is a purely instructional document (SKILL.md) that defines a set of verification rules (EVID-1 through EVID-10) for the agent to follow. It provides guidance on how to tag claims and verify information against production sources without providing any code to perform these actions automatically.
- [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to read external sources in full and verify data at production SHAs. This constitutes a data ingestion surface, but since the skill lacks executable capabilities or tools to perform file writes or network exfiltration, it does not pose a security risk in this context.
- Ingestion points: External sources, deployment records, and log files mentioned in verification rules.
- Boundary markers: Absent; the skill does not specify delimiters for untrusted content.
- Capability inventory: No scripts or executable tools are provided within the skill package.
- Sanitization: Not applicable as the skill does not process or transform data via code.
Audit Metadata