The Agent Skills Directory

[REMOTE_CODE_EXECUTION] (CRITICAL): The file references/cli.md instructs users to download a shell script from https://raw.githubusercontent.com/auth0/auth0-cli/main/install.sh and pipe it directly to the shell (| sh). This constitutes an unverified remote code execution pattern from a source not listed in the Trusted External Sources.
[CREDENTIALS_UNSAFE] (HIGH): Instructions in references/cli.md recommend the use of the auth0 apps show <app-id> --reveal command, which explicitly outputs sensitive client secrets in plain text to the terminal and agent context.
[COMMAND_EXECUTION] (MEDIUM): The framework detection logic in SKILL.md performs system commands like cat package.json | grep and ls -la | grep. These commands can be influenced by malicious file content or names in the local environment.
[EXTERNAL_DOWNLOADS] (LOW): The skill encourages downloading and installing the Auth0 CLI through third-party package managers (Homebrew, Scoop, Chocolatey).
[PROMPT_INJECTION] (LOW): Indirect Prompt Injection surface (Category 8) identified. 1. Ingestion: Reading package.json and directory listings. 2. Boundary markers: Absent. 3. Capability inventory: System command execution and CLI interaction. 4. Sanitization: Absent.

auth0-quickstart