content-newsletter-draft
Pass
Audited by Gen Agent Trust Hub on Aug 27, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests user-provided article outlines and topic ideas to generate a structured handoff for a downstream drafting pipeline. This creates a potential surface where malicious instructions within the user input could influence the behavior of the content-writer agent.
- Ingestion points: User outline or topic idea extracted in SKILL.md (Step 1).
- Boundary markers: The skill uses a structured Markdown template to delimit user content when handing off to the next agent.
- Capability inventory: The skill utilizes the google-workspace MCP to read and write to Google Drive and invokes internal pipeline tools like /zynkr-content-writer.
- Sanitization: No explicit escaping or sanitization of the user outline is specified before interpolation into the handoff prompt.
- [EXTERNAL_DOWNLOADS]: The skill references specific Google Drive folders and documents for style guides, forbidden words, and content history. These resources are used as context for the agent to ensure consistency with the established brand voice.
Audit Metadata