The Agent Skills Directory

[DATA_EXFILTRATION]: The core functionality of the skill is to upload local file content or stdin data to an external service (GitHub Gists). An agent could potentially be manipulated to upload sensitive files (e.g., .env, SSH keys, or internal documentation) to public or unlisted Gists.
[COMMAND_EXECUTION]: The script scripts/publish_gist.py uses the subprocess.run module to interact with the system environment.
It executes the gh CLI tool for gist creation and authentication status checks.
It uses system-specific commands (open, xdg-open, start) to open URLs in a web browser.
While the script generally uses argument lists to mitigate shell injection, the Windows implementation for opening URLs utilizes shell=True, which is a broader execution surface.
[CREDENTIALS_UNSAFE]: The skill attempts to access sensitive environment variables, specifically GITHUB_GIST_TOKEN and GITHUB_TOKEN, to authenticate API requests if the gh CLI is unavailable. While this is standard behavior for GitHub-integrated tools, it highlights the skill's capability to access and utilize account-level secrets.

github-gist