The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill specifically instructs the agent to ask the user for their Cloudflare Global API Key and account email. While the instructions advise using scoped tokens and caution against logging keys, requesting a Global API Key (which provides full account control) and using it within shell command templates increases the risk of credential exposure in environment logs or command history.
[COMMAND_EXECUTION]: The investigation methodology relies on executing bash commands and Python scripts that incorporate user-supplied parameters such as domain names. This pattern creates an attack surface for command injection where a malicious or poorly formatted input could be used to execute unauthorized commands on the host system.
[INDIRECT_PROMPT_INJECTION]: The skill processes both user-provided inputs and external JSON data from the Cloudflare API without implementing rigorous sanitization or explicit boundary markers. This exposes the agent to possible manipulation if the external data source or user inputs contain adversarial instructions. Ingestion points: User input fields (domain, email, api_key) and JSON responses from the Cloudflare client API. Boundary markers: Not present in the provided command templates or guidance. Capability inventory: Shell command execution via curl, dig, and openssl, as well as Python script execution using the requests library. Sanitization: The skill lacks logic to validate or escape user-provided strings before they are used in shell commands or prompt context.

cloudflare-troubleshooting