The Agent Skills Directory

[COMMAND_EXECUTION]: The skill is designed to manage an allowedCommands list within builder.config.json. This configuration typically dictates which shell commands the agent or environment is permitted to run. While the skill does not execute the commands itself, it serves as a management interface for the execution policy.
[SAFE]: The skill incorporates several security best practices:
User-in-the-loop: It uses AskUserQuestion to ensure that no changes are made to the configuration file without explicit user selection and consent.
Input Validation: The instructions explicitly forbid the inclusion of shell metacharacters (e.g., ;, &&, ||, |, backticks, $(, or newlines) in custom command entries, which mitigates risk of command injection via the configuration file.
Integrity Preservation: It is instructed not to remove existing entries or add duplicates, maintaining the integrity of the existing security policy.

allow-commands