headline-card
Pass
Audited by Gen Agent Trust Hub on Oct 5, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to fetch and process untrusted data from external sources (news articles and social media posts) to populate headline cards in a presentation.
- Ingestion points: External content is retrieved using tools (e.g., reader-list-documents) and stored in JSON and Typst files.
- Boundary markers: There are no explicit markers or instructions to treat external data as untrusted or to ignore instructions embedded within the text.
- Capability inventory: The skill involves file read/write operations and the use of command-line tools like grep for file verification.
- Sanitization: The skill lacks instructions for sanitizing or validating the external text before it is used in the layout.
Audit Metadata