The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill uses git clone and the GitHub API (gh api) to download content from the external repository Oruga420/claude-code-skills. These operations fetch unvetted data into the local environment.
[REMOTE_CODE_EXECUTION]: The skill provides automated workflows to download remote files and move them into the ~/.claude/skills/ directory. Since skills can contain executable components or complex instructions, this mechanism facilitates the persistent installation of remote code into the agent's runtime environment.
[COMMAND_EXECUTION]: The skill executes various shell commands to manage the local environment and interact with GitHub, including ls, git clone, cp, and gh api.
[PROMPT_INJECTION]: The skill presents an indirect prompt injection surface by ingesting and applying remote content directly into the agent's context without sanitization or boundary markers.
Ingestion points: GitHub API (gh api) and Git (git clone) are used to fetch content from Oruga420/claude-code-skills.
Boundary markers: None identified. Content fetched via the API is decoded and applied directly to the conversation.
Capability inventory: The skill has access to shell execution (ls, git clone, cp, gh api) and file system manipulation within the agent's configuration folders.
Sanitization: No sanitization or validation of the remote skill content is performed before application or installation.

skill-shed