export default async function main(args) { const { openaiApiKey, chatResponse } = args.inputVars; if (!chatResponse || !openaiApiKey) { return { next: { path: 'error' }, trace: [{ type: "debug", payload: { message: "Missing required input variable: chatResponse or openaiApiKey" } }] }; } const url = `https://api.openai.com/v1/chat/completions`; const domain = "pediatric dentistry and children's dental health"; const dental_health_criteria = ` Assess the presence of misinformation, harmful advice, or non-related content in the context of pediatric dentistry and children's dental health. The content should: 1. Contain accurate and safe advice about pediatric dentistry and children's dental health. 2. Be relevant to pediatric dentistry and children's dental health.`; const dental_health_steps = ` 1. Read the content and the criteria carefully. 2. Assess how much misinformation, harmful advice, or non-related content is contained in the content. 3. Assign a dental health score from 1 to 5, with 1 being no misinformation, harmful advice, or non-related content, and 5 being multiple instances of misinformation, harmful advice, or non-related content.`; async function checkOutputGuardrail() { const moderationSystemPrompt = `You are a moderation assistant. Your role is to detect content about {domain} in the text provided, and mark the severity of that content. ## {domain} ### Criteria {scoring_criteria} ### Instructions {scoring_steps} ### Content {content} ### Evaluation (score only!)`; const modMessages = [ { "role": "user", "content": moderationSystemPrompt.replace('{domain}', domain) .replace('{scoring_criteria}', dental_health_criteria) .replace('{scoring_steps}', dental_health_steps) .replace('{content}', chatResponse) }, ]; const response = await fetch(url, { method: 'POST', headers: { 'Content-Type': 'application/json', 'Authorization': `Bearer ${openaiApiKey}` }, body: JSON.stringify({ model: "gpt-4o", messages: modMessages, temperature: 0 }) }); if (!response.ok) { throw new Error('HTTP error! status: ' + response.status); } const data = await response.json; return data.choices[0].message.content; } try { const moderationResponse = await checkOutputGuardrail(); const score = parseInt(moderationResponse, 10); let message; if (score >= 3) { message = `Moderation guardrail flagged with a score of ${score}.`; return { next: { path: 'moderation_triggered' }, trace: [{ type: "debug", payload: { message } }] }; } else { message = "Passed moderation."; return { next: { path: 'continue' }, trace: [{ type: "debug", payload: { message } }] }; } } catch (error) { return { next: { path: 'error' }, trace: [{ type: "debug", payload: { message: `Error in moderation guardrail: ${error.message}` } }] }; }}