Neosantara + Hitori WA Bot Plugin Guide

Building a WhatsApp Bot is no longer a hassle. With the help of the Neosantara AI Gateway, you can turn your WhatsApp bot (e.g., Hitori Bot) into a super-smart assistant that's not only great at chatting but also understands our context and can think logically (reasoning).

In this article, we'll discuss the easy way to add the Neosantara plugin to the Hitori bot base. Check it out!

#What Do You Need Before Starting?

Before getting into the technicalities, make sure you already have the Hitori bot base. Hitori Bot is one of the best WA bot bases from Nazedev. The base is tidy, easy to modify, and has tons of features.

If you don't have it yet, just clone the official repo here: 👉 GitHub: nazedev/hitori

Once it's cloned and running, then proceed to the integration steps below.

#Why Use Neosantara?

Neosantara is a fast gateway that gives you access to the latest AI models (LLMs). What are the advantages?

1. OpenAI Compatible: No need to worry about learning new ways. If you've used the OpenAI API, the usage is exactly the same!
2. Archipelago 70B: This is a special model that is "very Indonesian". It understands context, culture, and even our slang.
3. DeepSeek R1: A model that's great at thinking. It will "reflect" first through chain-of-thought before giving an answer.
4. Local Gateway: The server is close, so the response is fast with minimal delay.

#How Do You Set Up Your API Key?

You need a "key" to open access:

1. Open Neosantara Dashboard.
2. Register and then verify your email to get a free balance for testing.
3. Create a new API Key, then copy and save it. Don't lose it!

Open the settings.js file in your Hitori bot folder. Find the global.APIKeys section and add the Neosantara URL plus the key you just created.

// settings.js

global.APIKeys = {
	'https://api.naze.biz.id': 'nz-xxxxxx',
	'https://api.neosantara.xyz/v1': 'YOUR_NEOSANTARA_API_KEY_HERE' // Add this!
}

Already have your API key? Sign up for Neosantara if you haven't yet — free Rp 10,000 credit, no credit card needed. Then continue below.

#How Do You Add AI Commands to naze.js?

In Hitori, all commands usually gather in naze.js using the switch(command) system. We'll add two new commands: .archipelago for casual chat and .r1 for heavy questions that require thinking.

Find switch(command) in naze.js, then paste this code:

// naze.js

case 'archipelago': {
    if (!text) return m.reply("Hello! How can I help you today?")
    m.reply(mess.wait)
    
    try {
        const response = await axios.post('https://api.neosantara.xyz/v1/chat/completions', {
            model: "archipelago-70b",
            messages: [{ role: "user", content: text }]
        }, {
            headers: {
                'Authorization': `Bearer ${global.APIKeys['https://api.neosantara.xyz/v1']}`,
                'Content-Type': 'application/json'
            }
        })
        
        await m.reply(response.data.choices[0].message.content)
    } catch (e) {
        console.error(e)
        m.reply("Oops, there was a problem asking Neosantara.")
    }
}
break

case 'r1': {
    if (!text) return m.reply("Try asking a hard question, I'll solve it!")
    m.reply("Wait a bit, thinking... 🧠")
    
    try {
        const response = await axios.post('https://api.neosantara.xyz/v1/chat/completions', {
            model: "deepseek-r1",
            messages: [{ role: "user", content: text }],
            thinking: { type: "enabled", budget_tokens: 2048 } // This is the secret to its intelligence
        }, {
            headers: {
                'Authorization': `Bearer ${global.APIKeys['https://api.neosantara.xyz/v1']}`,
                'Content-Type': 'application/json'
            }
        })
        
        const result = response.data.choices[0].message;
        // If there's a thinking process (reasoning), we display it too
        const thought = result.reasoning_content ? `*Thinking Process:*
_${result.reasoning_content}_

` : '';
        await m.reply(thought + result.content)
    } catch (e) {
        console.error(e)
        m.reply("Failed to get an answer from DeepSeek R1. Try again later.")
    }
}
break

#Can You Try Other Models?

The cool thing about the Neosantara AI Gateway is that there are many choices. You just need to change the model content in the code earlier to try other sensations:

• Gemini 3 Flash Preview: Most intelligent for matters requiring lightning-fast response.
• Garda Beta Mini: Very light, suitable for bots that don't want to be complicated.
• Kimi K2: Great for reading documents or long chats because its context window is wide.

Just change the name, and your bot instantly has a new "brain" with a different character!

#Conclusion

It's very easy to integrate Neosantara into Hitori, right? It only takes a few lines of code, and your WA bot has already moved up to become a super-advanced AI.

Happy exploring, don't forget to show off your new bot to your friends!

#Frequently Asked Questions

#What if I don't have a Hitori bot base yet?

Clone the official Hitori repo from GitHub: nazedev/hitori. The base is tidy, well-documented, and easy to modify. Once it's running, follow the steps above to add Neosantara.

#Which model should I start with?

Start with DeepSeek R1 for best reasoning quality, or Garda Beta Mini if you need the fastest response times for chat. You can switch models at any time by changing the model parameter — no code changes needed.

#Does Neosantara work outside Indonesia?

Yes. Neosantara's gateway is optimized for ASEAN with local infrastructure in Indonesia, but works globally. Response times may vary by region, but all models and features are fully accessible from anywhere.

#How much does this cost?

Neosantara offers Rp 10,000 free credit on signup — enough to build and test your bot extensively. After that, pay-as-you-go with Rupiah billing. No minimum commitments.

#References

• Neosantara Official Documentation
• Hitori Bot (Nazedev) — WhatsApp bot base used in this tutorial
• Neosantara GitHub Organization
• How to Use Any LLM with Neosantara — Guide to switching between AI models
• LiteLLM Integration Guide — Unified API for multiple LLM providers

Article published: June 20, 2026. Links, API endpoints, and model availability verified as of the publication date.