The Agent Skills Directory

COMMAND_EXECUTION (SAFE): The skill provides utility shell scripts (check-versions.sh and local-dev-setup.sh). These are local developer tools used for checking dependency versions and configuring environment variables. They do not download or execute remote code and do not perform any privileged operations.
CREDENTIALS_UNSAFE (SAFE): The local setup script (local-dev-setup.sh) handles database credentials by prompting for them securely and saving them to a .env.local file, which it automatically adds to .gitignore to prevent accidental exposure.
PROMPT_INJECTION (SAFE): The skill's documentation and metadata are focused on technical implementation and do not contain instructions that attempt to override AI safety filters or manipulate agent behavior.
DATA_EXFILTRATION (SAFE): The provided code templates strictly use Cloudflare Hyperdrive bindings for database connectivity and do not contain any patterns suggesting the exfiltration of sensitive data to external or untrusted domains.
DYNAMIC_EXECUTION (SAFE): The skill explicitly warns against the use of eval() in the mysql2 driver and provides the necessary configuration (disableEval: true) to comply with the secure execution environment of Cloudflare Workers.

cloudflare-hyperdrive