// ---------- Imports ----------
import { z } from "zod";
import { Prompt } from "../core";
// ---------- Schemas ----------
/**
* Arguments schema for removing LLM-specific language patterns
*/
const DegptContentArgsSchema = z.object({
content: z.string().describe("Text content to process"),
});
type DegptContentArgs = z.infer<typeof DegptContentArgsSchema>;
/**
* Prompt to remove common LLM/GPT language patterns
*/
export const prompt = new Prompt<DegptContentArgs>(
{
name: "degpt-content",
description:
"Remove common LLM/GPT language patterns and mannerisms from text",
argsSchema: DegptContentArgsSchema,
},
async (args) => {
return [
{
role: "user",
content: {
type: "text",
text: `Review and revise this text to remove common AI language patterns like:
- Phrases that start with "This isn't X, it's Y"
- Marketing buzzwords like "game changer", "revolutionary", "groundbreaking"
- Overly enthusiastic or artificial-sounding language
- Repetitive acknowledgments and confirmations
- Unnecessarily formal or robotic transitions
Here is the text to process:
${args.content}
Provide the revised text with natural, straightforward language.`,
},
},
];
},
);
