monitoring-config-auditor
Pass
Audited by Gen Agent Trust Hub on Feb 16, 2026
Risk Level: LOWPROMPT_INJECTION
Full Analysis
- [Indirect Prompt Injection] (LOW): The skill ingests untrusted infrastructure code, which could contain instructions to bypass audit rules. * Ingestion points: Files within the user-provided 'dir' path as defined in SKILL.md. * Boundary markers: None specified in the provided documentation. * Capability inventory: Local file read and write access through the 'dir' and 'out' arguments; no subprocess or network calls identified in the manifest. * Sanitization: No evidence of sanitization for incoming configuration data prior to processing.
- [Unverifiable Dependencies] (SAFE): The project uses a workspace-managed dependency (@agent/core), which is a trusted internal reference within the agent repository.
Audit Metadata