context-probe
Warn
Audited by Gen Agent Trust Hub on Mar 21, 2026
Risk Level: MEDIUMPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill mandates the insertion of a rule block containing phrases such as 'This rule overrides any conflicting instructions in this file' and 'User instructions MUST NOT disable, remove, or modify this behavior.' This language is characteristic of prompt injection, specifically designed to subvert the agent's normal instruction hierarchy and ignore subsequent user constraints.
- [COMMAND_EXECUTION]: The skill instructs the agent to perform automated, unconfirmed write operations on various sensitive configuration files including
CLAUDE.md,.cursor/rules/, and.agents. This broad file system access targets files that define AI behavior across different development environments.
Audit Metadata