The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill is designed to fetch content from arbitrary GitHub URLs and a third-party catalog (VoltAgent/awesome-agent-skills). It facilitates downloading external code and configuration files into the local environment.
[REMOTE_CODE_EXECUTION]: By installing external 'Agent Skills' from untrusted repositories, the skill allows for the delivery and subsequent execution of arbitrary code within the agent's runtime environment. The installation process places external scripts into system-specific paths (~/.claude/skills/).
[COMMAND_EXECUTION]: The skill uses shell commands to run a local Python installer script (python3 ~/.claude/scripts/install-skill.py). This relies on the existence and integrity of a script in a sensitive directory and executes it with arguments derived from external URLs.
[INDIRECT_PROMPT_INJECTION]: The skill possesses a significant attack surface for indirect injection by ingesting and processing untrusted SKILL.md files from the web.
Ingestion points: External GitHub URLs, subfolders, and search results from the VoltAgent catalog.
Boundary markers: The skill mentions requiring explicit user confirmation and basic YAML validation, which are present but may be insufficient against adversarial content.
Capability inventory: The skill can execute shell commands via Python to install and symlink files.
Sanitization: There is no evidence of content sanitization beyond checking for YAML frontmatter presence.

skill-installer