The Agent Skills Directory

[COMMAND_EXECUTION]: The skill contains a 'Companion check' directive in SKILL.md that instructs the agent to execute shell commands (ls) against several local and home directory paths, including ~/.claude/skills/, ~/.agent/skills/, and ~/.agents/skills/. This probes the host filesystem for environment-specific information about installed AI agent extensions.
[EXTERNAL_DOWNLOADS]: The documentation and skill logic promote the use of npx skills add AbsolutelySkilled/AbsolutelySkilled. This command utilizes the npx package runner to download and execute code from an external, non-whitelisted source, which represents a potential remote code execution vector if the repository or the 'skills' package is untrusted or compromised.
[PROMPT_INJECTION]: The skill includes behavioral overrides, such as requiring the agent to start responses with a specific emoji (🧢) and mandating a 'Companion check' on first activation. This check forces the agent to perform system discovery and output specific installation commands, steering the agent's behavior to promote other skills based on the user's local environment state.

system-design