The Agent Skills Directory

[COMMAND_EXECUTION]: The scripts/install-skill-from-github.py script uses subprocess.run() to execute git commands for cloning and sparse-checking repositories. This is the intended primary purpose of the skill and is implemented with argument lists (avoiding shell interpolation). While the static analyzer flagged this as a potential risk, review confirms it is used for legitimate repository management operations.
[EXTERNAL_DOWNLOADS]: The skill downloads content from github.com and codeload.github.com. These are well-known services and the default repository (openai/skills) is from a trusted organization. The downloads are used to fetch skill metadata and source files as requested by the user.
[DATA_EXPOSURE]: The skill accesses environment variables GITHUB_TOKEN or GH_TOKEN to authenticate API requests to GitHub. This is a standard practice for managing credentials and no evidence was found of these tokens being exfiltrated or logged improperly.
[SECURITY_BEST_PRACTICES]: The _safe_extract_zip function in scripts/install-skill-from-github.py implements validation to ensure files are not extracted outside the intended destination directory, preventing ZipSlip/directory traversal attacks. The skill also validates that skill names and paths are relative and do not contain parent directory markers (..).

skill-installer