copywriting
Pass
Audited by Gen Agent Trust Hub on Oct 11, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONNO_CODE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructions direct the agent to load and process data from an external local file to define target personas and brand voice, creating a surface for indirect instructions.
- Ingestion points: The skill explicitly references
sales-marketing.local.mdin Phase 1 (Brief Collection) and as a mandatory constraint in the "NEVER DO THESE" section. - Boundary markers: The prompt instructions lack explicit delimiters or warnings to treat the content of the loaded file as untrusted data, increasing the likelihood that the agent will follow instructions embedded within that file.
- Capability inventory: The skill is designed to generate multiple variants of marketing copy (emails, ads, landing pages) based on the input, which provides a high-volume vector for propagating instructions from the source file.
- Sanitization: There are no instructions provided to sanitize, filter, or validate the content ingested from the external markdown file.
- [NO_CODE]: The skill consists entirely of instructional markdown and evaluation configurations, with no accompanying scripts, executables, or binary dependencies, which reduces the overall technical risk profile.
Audit Metadata