The Agent Skills Directory

EXTERNAL_DOWNLOADS (MEDIUM): The file scripts/check_install_cli.sh executes npm install -g swiftopenai-cli. This downloads and installs a package from the public npm registry. Since the author and package are not on the trusted source list, this is classified as an unverifiable dependency installation.
COMMAND_EXECUTION (LOW): The scripts execute several local system commands including npm, swiftopenai, tr, and cut to manage the installation and configuration of the CLI tool.
PERSISTENCE_MECHANISMS (LOW): The script scripts/configure_provider.sh and the documentation in references/providers.md encourage users to append API keys to shell profile files such as ~/.zshrc or ~/.bashrc. Although the script does not automate this modification, it explicitly suggests a pattern for persisting sensitive environment variables.
CREDENTIALS_UNSAFE (SAFE): The skill does not contain hardcoded secrets. It includes placeholders (e.g., sk-..., gsk_...) and directs users to provide their own keys via environment variables or interactive configuration.

llm-router