suggesting-cursor-rules
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill facilitates the creation of persistent rule files based on user-provided feedback and corrections. This represents a potential vulnerability surface where malicious input could be used to permanently alter agent behavior for a specific workspace.
- Ingestion points: User conversation history, specific corrections, and repeated instructions (as defined in the 'Triggers' and 'How to Suggest' sections of SKILL.md).
- Boundary markers: Absent. There are no guidelines or delimiters provided to ensure the agent differentiates between valid stylistic preferences and adversarial prompt injections.
- Capability inventory: The skill utilizes file system write capabilities to create markdown (.mdc) files within the .cursor/rules/ directory.
- Sanitization: Absent. The content of the rule is derived directly from the user's conversational input without explicit validation or safety filtering instructions.
Audit Metadata