skill-collab-proposal
Pass
Audited by Gen Agent Trust Hub on Sep 2, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from external 'brand briefs' and 'supplementary materials' to generate proposals, creating a surface for potential injection attacks where embedded instructions could attempt to influence agent behavior.
- Ingestion points: Step 1 in
SKILL.mdidentifies '品牌 brief' and '补充材料' as optional inputs that are extracted for processing. - Boundary markers: No explicit delimiters or specific instructions are defined to isolate this external content or to prevent the agent from following instructions contained within those documents.
- Capability inventory: The skill includes the capability to write output files to the
outputs/directory as described in Step 9 of the execution flow. - Sanitization: There are no described mechanisms for sanitizing, filtering, or escaping the content of the provided briefs before interpolation into the generation process.
Audit Metadata