The Agent Skills Directory

[PROMPT_INJECTION]: The skill employs strong, coercive language to force specific behaviors and override the agent's logic.
Evidence in SKILL.md: "If you think there is even a 1% chance a skill might apply to what you are doing, you ABSOLUTELY MUST invoke the skill."
Evidence in SKILL.md: "IF A SKILL APPLIES TO YOUR TASK, YOU DO NOT HAVE A CHOICE. YOU MUST USE IT. This is not negotiable. You cannot rationalize your way out of this."
The "Red Flags" section explicitly instructs the agent to ignore its own internal reasoning (e.g., thinking a question is "simple" or needing "more context first") when deciding whether to follow its default behavior or the skill's instructions.
[COMMAND_EXECUTION]: The reference files contain shell script snippets and tool definitions that involve command execution on the host system.
Evidence in references/codex-tools.md: Provides git commands for environment detection such as GIT_DIR=$(cd "$(git rev-parse --git-dir)" 2>/dev/null && pwd -P).
Evidence in references/gemini-tools.md: Explicitly maps actions to tools including run_shell_command, grep_search, and web_fetch.

using-superpowers