// as the tool will be public, we need some level of moderation to bred unfair use | |
// note: another possibility could be to use the moderation api from openai, | |
// but it is paying | |
const config = JSON.parse(`${process.env.VC_MODERATION_FILTER_PATTERN}`) | |
export const hasBadWords = (userPrompt: string) => { | |
const normalized = userPrompt.trim().toLowerCase() | |
return !!config[normalized] | |
} |