The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The documentation references commands like glab orbit and glab duo cli that download and execute binaries from GitLab infrastructure. It also includes instructions for installing the tool via third-party repositories, such as executing a script from upciti/wakemeops using curl | sudo bash.
[PERSISTENCE]: The documentation provides instructions for configuring shell completions for various shells, which involve appending load commands to shell profile files like ~/.zshrc.
[PRIVILEGE_ESCALATION]: The documentation describes operations requiring elevated permissions, such as system-wide completion installation in /etc/bash_completion.d/ and granting filesystem access to the system keychain for authentication in sandboxed environments.
[DYNAMIC_EXECUTION]: The documentation explains the valueFromCommand configuration setting, which allows the glab CLI to execute user-defined shell commands to generate dynamic HTTP header values.
[METADATA_POISONING]: The skill metadata labels the content as the "Official command reference for glab," although the skill is maintained by a community contributor ('wenerme') rather than the official GitLab organization.

glab-docs