QuickStart Guide

ChainGPT Web3 AI Chatbot & LLM – QuickStart Guide (Tech‑Team‑Verified)

This short guide shows you how to call ChainGPT’s Web3 LLM in under five minutes—either with plain HTTPS or the official JavaScript SDK. Everything here mirrors the internal Tech‑Team reference, not the older public docs.

---

1. Prerequisites ✔︎

What you need	Notes
ChainGPT account + API key	Create a key in AI Hub → API Dashboard and copy it once.
Credits	Each request = 0.5 credits.Turning chatHistory on adds +0.5 credits per call
HTTP client	cURL, Postman, fetch, axios, etc.
(If using SDK) Node ≥ 14	Install from npm.
Secure key storage	Put the key in an env var, secret manager, or server config—never ship it to the browser.

Tip export CHAINGPT_API_KEY="sk‑***" so the examples just work.

---

2. Integration Options

Option	When to choose	How it works
REST API	Any language / server	POST https://api.chaingpt.org/chat/stream (single endpoint for blob and streaming)
JavaScript SDK	Node or a build‑step web app	npm install @chaingpt/generalchat → call createChatBlob() or createChatStream()

---

3. QuickStart via REST API

3.1 Authentication & Endpoint

Authorization: Bearer $CHAINGPT_API_KEY
Content-Type: application/json
POST https://api.chaingpt.org/chat/stream

There is no separate /chat endpoint—all chat traffic goes to /chat/stream.

3.2 Single‑Shot (JSON “blob”) response

curl -X POST https://api.chaingpt.org/chat/stream \
  -H "Authorization: Bearer $CHAINGPT_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
        "model":"general_assistant",
        "question":"Hello, ChainGPT! Who are you?",
        "chatHistory":"off"
      }'

If you don’t treat the response as a stream, cURL (or your HTTP client) waits until the LLM finishes, then returns one JSON:

{ "status":true,
  "data": { "bot": "Hello! I’m ChainGPT, a Web3‑savvy AI assistant…" } }

3.3 Streaming in real time

curl -N -X POST https://api.chaingpt.org/chat/stream \
  -H "Authorization: Bearer $CHAINGPT_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
        "model":"general_assistant",
        "question":"Give me the latest ETH stats.",
        "chatHistory":"off"
      }'

-N turns off buffering so you see partial chunks instantly. Concatenate chunks on arrival until the connection closes.

3.4 Conversation Memory (chat history)

curl -X POST https://api.chaingpt.org/chat/stream \
  -H "Authorization: Bearer $CHAINGPT_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
        "model":"general_assistant",
        "question":"Remember my token ABC and what it does.",
        "chatHistory":"on",
        "sdkUniqueId":"user42"
      }'

Subsequent calls with the same sdkUniqueId and chatHistory:"on" include the prior Q&A so the bot answers in context.

Cost impact: chatHistory:"on" consumes +0.5 credits per request.

3.5 Custom Context & Tone

{
  "question":"Explain our project.",
  "useCustomContext":true,
  "contextInjection":{
     "companyName":"ABC Crypto",
     "companyDescription":"A DeFi yield platform",
     "aiTone":"PRE_SET_TONE",
     "selectedTone":"FRIENDLY"
  }
}

Set useCustomContext:true; include only the fields you need. Omit contextInjection to fall back to the default context configured in AI Hub.

---

4. QuickStart via JavaScript SDK

4.1 Install & init

npm install @chaingpt/generalchat

# or 
yarn add @chaingpt/generalchat

import { GeneralChat } from "@chaingpt/generalchat";
const chat = new GeneralChat({ apiKey: process.env.CHAINGPT_API_KEY });

4.2 Blob response

const res = await chat.createChatBlob({
  question: "Hi, what is ChainGPT?",
  chatHistory: "off"
});
console.log(res.data.bot);

createChatBlob() buffers the /chat/stream response for you.

4.3 Streaming response

const stream = await chat.createChatStream({
  question: "Summarise the last BTC block.",
  chatHistory: "off"
});
for await (const chunk of stream) process.stdout.write(chunk);

4.4 Chat history & context

await chat.createChatBlob({
  question: "Track ABC token price.",
  chatHistory: "on",
  sdkUniqueId: "user42",
  useCustomContext: true
});

SDK parameters match the REST fields 1‑for‑1.

---

5. Error Codes — Quick Reference

Code	Meaning	Fix
401	Missing / bad API key	Check Authorization header.
402 / 403	Out of credits	Top‑up in AI Hub.
400	Bad JSON / missing field	Verify model and question.
5xx	Server error	Retry after brief delay.

---

6. Best Practices

• Hide the key. Never embed it in client JavaScript—proxy via your backend.

• Watch credits. One call = 0.5 credits; history = +0.5. Set alerts.

• Use unique sdkUniqueId. One per user/session keeps histories clean.

• Handle stream ‘end’ & ‘error’. Always close or retry gracefully.

• Stay updated. Check the npm changelog for new SDK features or model IDs.

---

You’re now ready to integrate ChainGPT using the exact endpoints and parameters defined by the Tech Team. For deeper dives, see the full API Reference or SDK docs. Happy building!