subject-line
Pass
Audited by Gen Agent Trust Hub on Nov 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted content from local workspace directories and user-referenced files to provide context for AI generation.
- Ingestion points: Document intelligence logic searches the outreach/, campaigns/, marketing/, and emails/ directories for subject line documents and uses slash commands like /context-recall to pull data.
- Boundary markers: There are no instructions to use protective XML tags or delimiters to isolate untrusted file content from the system prompt instructions.
- Capability inventory: The system allows reading existing files and writing generated subject lines to the local file system.
- Sanitization: The skill lacks validation steps to ensure that the content within retrieved marketing files does not contain prompt injection or malicious instructions that could manipulate the output or behavior.
Audit Metadata