qws-zoom-out
Pass
Audited by Gen Agent Trust Hub on Nov 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection because it is designed to ingest and analyze potentially untrusted user content.
- Ingestion points: As described in
SKILL.mdandREADME.md, the agent reads 'current conversation', 'provided materials', and project-related files like 'account positioning documents' and 'content lists'. - Boundary markers: The instructions do not mandate the use of strict boundary markers (such as XML tags or unique random delimiters) to encapsulate untrusted data, which could allow instructions embedded in user files to influence the agent's logic.
- Capability inventory: The skill's potential impact is low because it lacks access to high-risk capabilities. It does not perform network operations, subprocess execution, or sensitive file system modifications.
- Sanitization: There are no instructions for the agent to sanitize or validate the content of ingested files for embedded malicious instructions.
Audit Metadata