email-marketing-ops
Pass
Audited by Gen Agent Trust Hub on Oct 23, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill establishes a workflow where agents are instructed to locate, read, and prioritize instructions from an external file named
brand-context.mdlocated in the user's project directories. - Ingestion points: The agent is directed to check the project root,
./.claude/brand-context.md, and./docs/brand-context.mdfor binding instructions. - Boundary markers: The skill lacks explicit boundary markers or safety warnings for this ingested content, instead instructing the agent to "treat it as binding" and that its rules "override an agent's own defaults."
- Capability inventory: The agents possess the authority to generate email copy, implement marketing automation workflows in platforms like HubSpot and Marketo, and configure sensitive domain authentication records (SPF, DKIM, DMARC).
- Sanitization: The skill emphasizes adherence to legal frameworks like GDPR and CAN-SPAM, but does not define sanitization or validation protocols for the instructions ingested from the context file itself, which could be used to steer agent behavior if the file contains adversarial content.
Audit Metadata