Issue
I'd love to use the detect urls feature, but am scared of the possibility of rogue URLs getting injected into the prompts somehow.
If we had a way to provide some sort of comma-separated list of allowed domains to let through, that would help make this a bit more secure perhaps.
Perhaps some sort of new flag like --allowed-domains <domain1>,[domain2,...] flag backed by an ENV CECLI_ALLOWED_DOMAINS=... or something better named.
Without it, my only option is to do some sort of URL redaction pass of the prompt message file to filter out domains i don't recognize before running cecli.
Version and model info
No response