li-recorder
Pass
Audited by Gen Agent Trust Hub on Sep 24, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill accepts untrusted user input (content ideas) and writes it directly to a local file, creating a surface for potential downstream injection if other skills process this file.
- Ingestion points: User topic ideas provided during the diagnostic workflow.
- Boundary markers: Absent; the skill instructions do not specify the use of delimiters when recording the text.
- Capability inventory: The skill has the capability to append text to a local markdown file (
01-内容生产/选题管理/00-选题记录.md). - Sanitization: Absent; the instructions explicitly command the agent to use the user's "original words" and "not polish" the text.
- [SAFE]: The identified indirect prompt injection surface is restricted to a specific local file path and is essential for the skill's primary function as a recorder. The skill does not perform any network operations, access sensitive system files, or execute external code.
Audit Metadata