email-ghostwriter
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONNO_CODE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted user-supplied data which could potentially influence the agent's behavior. \n
- Ingestion points: The skill instructions in SKILL.md direct the agent to use voice guides, examples, and facts provided by the user to draft emails. \n
- Boundary markers: The instructions do not specify the use of delimiters or 'ignore embedded instructions' warnings when interpolating this untrusted data into the prompt context. \n
- Capability inventory: SKILL.md explicitly mentions a 'sending action', indicating that the agent has access to tools for external communication. \n
- Sanitization: There are no requirements defined for sanitizing or filtering the user-provided data before it is processed. \n
- [NO_CODE]: The skill consists entirely of markdown instructions and template files without any associated scripts or executable code.
Audit Metadata