Implementation:Ucbepic Docetl ChatRoute

Knowledge Sources	Ucbepic_Docetl
Domains	Frontend, React_UI
Last Updated	2026-02-08 00:00 GMT

Overview

Concrete tool for the Next.js API route that provides AI chat functionality for the DocETL playground assistant.

Description

This API route handles chat requests from the AIChatPanel and PromptImprovementDialog components. It supports both Azure OpenAI and direct OpenAI backends, with automatic message truncation to stay within token limits (120K tokens / 500K characters). The route extracts the longest message for middle-section truncation, supports personal API keys via request headers, and tracks usage via Supabase. It conditionally sets temperature to 1 for GPT-5 models.

Usage

Called by the frontend chat components via useChat from the Vercel AI SDK. The route streams responses back to the client.

Code Reference

Source Location

• Repository: Ucbepic_Docetl
• File: website/src/app/api/chat/route.ts
• Lines: 1-191

Signature

export async function POST(req: Request): Promise<Response>

// Internal helper:
function truncateMessages(messages: ChatMessage[]): ChatMessage[]

Import

// This is an API route. Called via:
fetch("/api/chat", { method: "POST", body: JSON.stringify({ messages }) })

I/O Contract

Inputs (Props)

Name	Type	Required	Description
messages	ChatMessage[]	Yes	Array of chat messages with role and content
x-openai-key	string (header)	No	Personal OpenAI API key
x-use-openai	string (header)	No	Set to "true" to use OpenAI instead of Azure
x-namespace	string (header)	No	User namespace for tracking
x-source	string (header)	No	Source identifier (e.g., "ai_chat")

Outputs

Name	Type	Description
stream	ReadableStream	Streamed AI chat response

Usage Examples

const { messages, handleSubmit } = useChat({
  api: "/api/chat",
  headers: {
    "x-use-openai": "true",
    "x-openai-key": myApiKey,
    "x-namespace": "my-namespace",
    "x-source": "ai_chat",
  },
});

Related Pages

• Environment:Ucbepic_Docetl_Frontend_Node_Environment