brand-product-context
Pass
Audited by Gen Agent Trust Hub on Oct 22, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSNO_CODE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process information from untrusted external sources, such as websites and user-uploaded marketing materials, to create a persistent '.agents/product-marketing.md' brief.\n
- Ingestion points: Website URLs, pitch decks, marketing copy, and sales-call notes provided by the user (SKILL.md).\n
- Boundary markers: The instructions do not define strict boundary markers or include specific directives to ignore potential malicious instructions embedded within the source materials during extraction.\n
- Capability inventory: The skill utilizes file system read/write operations and network fetching (via agent tools) to populate the brand brief.\n
- Sanitization: No explicit sanitization or filtering of the extracted content is mentioned, which could allow instructions hidden in source documents to be persisted in the brand brief and influence future agent interactions.\n- [EXTERNAL_DOWNLOADS]: The skill instructs the agent to fetch and analyze content from user-provided URLs to gather brand facts.\n
- Evidence: The 'Workflow' section in SKILL.md explicitly directs the agent to 'Pull facts from the website'.\n- [NO_CODE]: The skill contains only Markdown instructions and templates, with no executable scripts or binaries included in the distribution.
Audit Metadata