API Assitant documatio v2

API
1

How can we develop and use the Assistans API when the current version is v2 but but the documention is for the v1 ?

I have tried to code my app to access Assistants I developed in the Playground. I have tried to use ChatGTP 4o to write the code and debug it. But now it seems that OpenAI LLMs are not trained to use and write code or advise on how to use the API and does not seam to understand what the v2 interface is.

1 Like
2

I have also defined my own GPTs at ChatGPT.
I have printed the API documentation as PDF files. Then, I uploaded these files to the GPTs.
I still do not get code advice that is working.
It seems that the documentation is written, so you need to have more skills than the AI to understand how the API interface works.

We have a saying, “The shoemaker’s children don’t have any shoes.” This seems to apply also here.

1 Like
3

I tried the same approach, but it didn’t work. I should mention that I’m completely new to coding. I have a question: When I build an Assistant in the Playground and upload some files, those files are stored in a vector store, and I receive an ID for that vector store. If I then send a request to my Assistant using its ID, will it automatically access the data in the vector store to provide the correct response?

Additionally, what about the prompt I set up for the Assistant? Does it play any role when I send an API request?

Lastly, how can I actually access the Assistant? I’ve created a project, an Assistant, and an API for the project. I’ve also added all the IDs to my .env file. However, I keep getting error messages like:

  • Error in the request to OpenAI: OpenAI API Error: Unrecognized request arguments supplied: assistant_id, vector_store_id
  • Error in the request to OpenAI: OpenAI API Error: Invalid URL (POST /v1/assistants/asst_nY7siE1iEtnXA8PtkZRWTflR/messages)
  • Error from OpenAI: Unknown parameter: 'assistant_id'.

These errors vary depending on how I try to access the Assistant.

This is my last version:
// Log-Ausgabe zum ĂśberprĂĽfen des Serverstarts
console.log(“Server startet…”);

// Import der benötigten Module
import express from “express”;
import bodyParser from “body-parser”;
import fetch from “node-fetch”;
import dotenv from “dotenv”;
import cors from “cors”;

// Laden der Umgebungsvariablen
dotenv.config();

// Initialisieren der Express-App
const app = express();
console.log(“Express-App initialisiert.”);

// Middleware-Konfiguration
app.use(express.static(“public”)); // Statische Dateien bereitstellen
app.use(bodyParser.json()); // JSON-Parsing
app.use(cors()); // CORS aktivieren

// POST-Endpunkt fĂĽr Chat-Anfragen
app.post(“/chat”, async (req, res) => {
const userMessage = req.body.message;

if (!userMessage) {
return res.status(400).json({ error: “Nachricht fehlt im Request-Body.” });
}

try {
// Anfrage an den Assistants-v2-Endpunkt mit Vector Store und Organisation
const response = await fetch(
https://api.openai.com/v2/assistants/${process.env.ASSISTANT_ID}/messages,
{
method: “POST”,
headers: {
“Content-Type”: “application/json”,
Authorization: Bearer ${process.env.OPENAI_API_KEY},
“OpenAI-Organization”: process.env.OPENAI_ORGANIZATION_ID, // Organisation hinzufügen
},
body: JSON.stringify({
message: {
role: “user”,
content: userMessage, // Nachricht des Nutzers
},
vector_store_id: process.env.VECTOR_STORE_ID, // Vector Store einbinden
}),
}
);

if (!response.ok) {
  const errorData = await response.json();
  throw new Error(
    `OpenAI-API-Fehler: ${errorData.error?.message || "Unbekannter Fehler"}`
  );
}

const data = await response.json();
res.json({ reply: data.message.content });

} catch (error) {
console.error(“Fehler bei der Anfrage an OpenAI:”, error.message);
res.status(500).json({ error: "Serverfehler: " + error.message });
}
});

// Definieren des Ports und Starten des Servers
const PORT = process.env.PORT || 3000;

console.log(“Starte Server…”);

try {
app.listen(PORT, () => {
console.log(Server läuft auf http://localhost:${PORT});
}).on(“error”, (err) => {
console.error(“Server konnte nicht gestartet werden:”, err.message);
});
} catch (err) {
console.error(“Fehler beim Starten des Servers:”, err.message);
}

// Globale Fehlerbehandlung
process.on(“uncaughtException”, (err) => {
console.error(“Ungefangene Ausnahme:”, err.message);
});

process.on(“unhandledRejection”, (reason) => {
console.error(“Unbehandelte Promise-Ablehnung:”, reason);
});

Error: Fehler bei der Anfrage an OpenAI: OpenAI-API-Fehler: Unknown request URL: POST /v2/assistants/asst_nY7siE1iEtnXA8PtkZRWTflR/messages. Please check the URL for typos, or see the docs at https://platform.openai.com/docs/api-reference/.

One last question: Will the Assistant feature be deprecated in the future, or will it remain available?