Implementation:Promptfoo Promptfoo Prompt Suggestions

Knowledge Sources	Promptfoo_Promptfoo
Domains	Prompt_Engineering, LLM_Integration
Last Updated	2026-02-14 07:45 GMT

Overview

Concrete tool for generating prompt variants by calling an LLM to suggest alternative phrasings of a given prompt.

Description

The Prompt_Suggestions module (suggestions.ts) provides the generatePrompts() function that uses the default OpenAI suggestions provider to create prompt variants. Given an input prompt, it sends it to the LLM with a system message requesting variants, and returns the generated alternatives along with token usage statistics.

Usage

Called from the web UI or CLI when users request automatic prompt suggestions or improvements.

Code Reference

Source Location

Repository: Promptfoo_Promptfoo
File: src/suggestions.ts
Lines: 1-50

Signature

export async function generatePrompts(
  prompt: string,
  _num: number,
): Promise<GeneratePromptsOutput>

Import

import { generatePrompts } from './suggestions';

I/O Contract

Inputs

Name	Type	Required	Description
prompt	string	Yes	The original prompt to generate variants for
_num	number	Yes	Number of variants requested (currently returns 1)

Outputs

Name	Type	Description
prompts	string[]	Array of generated prompt variants
error	string	Error message if generation failed
tokensUsed	TokenUsage	Token consumption statistics

Usage Examples

import { generatePrompts } from './suggestions';

const result = await generatePrompts('Summarize the following text: {{text}}', 1);

if (result.prompts) {
  console.log('Suggested variant:', result.prompts[0]);
}
console.log('Tokens used:', result.tokensUsed);

Related Pages

Page Connections

Double-click a node to navigate. Hold to expand connections.

Principle

Implementation

Heuristic

Environment