checkbox-picker
Pass
Audited by Gen Agent Trust Hub on Sep 27, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes system commands, such as
claude plugin uninstall, to manage installed components based on user interaction with the generated interface. - [INDIRECT_PROMPT_INJECTION]: The skill reads decision data from a JSON file in the
~/Downloads/directory and performs actions based on that data, which could be manipulated to trigger unintended operations. - Ingestion points: Data is read from files in
~/Downloads/during the execution phase (SKILL.md). - Boundary markers: None identified; the skill assumes the content of the downloaded file is a valid and safe set of identifiers.
- Capability inventory: Includes file system moves (
shutil.move), directory traversal, and CLI-based plugin uninstallation. - Sanitization: No validation or sanitization of the JSON structure or the identifiers within the file is specified before execution.
- [EXTERNAL_DOWNLOADS]: The skill fetches metadata from external sources, including the GitHub API, to populate the list of items in the picker interface.
Audit Metadata