next
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill analyzes local files such as brand-config.json and user-generated content in the drafts/ folder to determine project status. This represents an indirect prompt injection surface area where content in these files could attempt to influence the agent's logic.
- Ingestion points: brand-config.json, gtm/ directory, and drafts/ directory (SKILL.md).
- Boundary markers: The skill does not define specific delimiters to isolate external data from instructions.
- Capability inventory: Read Glob (allowed-tools).
- Sanitization: No content sanitization is performed on the ingested data.
- [EXTERNAL_DOWNLOADS]: The skill references an external plugin registry (@gtm-operator-skills) and provides users with installation commands for missing tools. This is a standard instructional pattern for modular toolsets.
- [TIME_DELAYED_CONDITIONAL]: The logic checks for the existence of recent files (e.g., posts from the last 7 days) to determine if a workflow step is complete. This is a benign functional requirement for status reporting.
Audit Metadata