gritql
Pass
Audited by Gen Agent Trust Hub on Sep 23, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill teaches the agent how to process and transform user-provided code, which introduces a surface for indirect prompt injection if the code being refactored contains malicious instructions.\n
- Ingestion points: User-provided code snippets and transformation requests (SKILL.md).\n
- Boundary markers: No explicit instructions provided to the agent to treat input code as data or to ignore embedded instructions.\n
- Capability inventory: Structural transformation of code, file system write operations via
$new_files, and custom logic execution (references/examples.md, references/syntax.md).\n - Sanitization: There are no defined protocols for validating or sanitizing the input code before the agent generates transformation patterns.\n- [DYNAMIC_EXECUTION]: The documentation describes the ability to define and use custom JavaScript functions within GritQL patterns to perform complex transformations (references/functions.md). While a core feature, this capability allows for dynamic logic execution during the refactoring process.\n- [COMMAND_EXECUTION]: The skill provides examples of using the
$new_filesglobal variable to programmatically create and write content to new files on the disk, which is a standard part of multi-file refactoring workflows (references/examples.md, references/syntax.md).
Audit Metadata