The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill directs users to install software extensions from external GitHub repositories (Q00/ouroboros and supercent-io/skills-template) that are not recognized as trusted vendors or well-known services.
[REMOTE_CODE_EXECUTION]: Setup instructions include the use of npx skills add with a remote repository URL, which involves downloading and executing third-party code. Additionally, the skill architecture for platforms like Claude Code and Gemini CLI relies on installing remote plugins and extensions that execute logic within the agent's environment.
[COMMAND_EXECUTION]: The provided setup-codex-hook.sh script executes shell commands to modify the local ~/.codex/config.toml file, injecting persistent command contracts into the developer instructions. The installation process for Gemini CLI also involves configuring AfterAgent hooks that execute a shell script (ralph-check.sh) after every response, establishing a persistent command execution loop.
[PROMPT_INJECTION]: The skill implements a persistent 'Ralph' mode that instructs the agent to continue iterating autonomously until internal verification passes. This logic is designed to override standard interaction boundaries and session constraints by enforcing a continuous execution loop.

ralph