byted-recruitment-jd-optimization
Pass
Audited by Gen Agent Trust Hub on Sep 21, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection and potential path traversal because it processes untrusted user data to determine file storage locations and content.
- Ingestion points: Untrusted data enters the agent's context through user-provided recruitment requirements and draft job descriptions, as described in the '工作流程' (Workflow) section of
SKILL.md. - Boundary markers: There are no specific instructions or delimiters provided to the agent to distinguish between its internal instructions and the content provided by the user, increasing the risk that the agent might obey instructions embedded within a draft JD.
- Capability inventory: The skill explicitly instructs the agent to perform file system write operations. It defines a save path based on the user-supplied job title:
jd-process/[职位名]/[职位名]-jd.md. - Sanitization: The instructions lack any requirement for the agent to sanitize the
[职位名](Job Title) variable. An adversary could provide a job title containing path traversal characters (e.g.,../../) to attempt to write files outside the intended workspace directory.
Audit Metadata