# Interview Coach Bot Workflow

> n8n mein Telegram pe apna AI interview coach: workflow, prompt aur setup step by step

Free guide by bloopcommits (@bloopcommits) · https://bloopcommits.pages.dev/coach/

Interview se pehle practice karwane wala koi nahi milta? Is guide mein ek chhota n8n workflow hai: Telegram bot ko topic bhejo (HR ya Python), bot ek sawaal poochta hai; tum us sawaal pe REPLY karke apna jawab bhejte ho, aur AI ek achhi cheez aur ek sudhaar ka tip deta hai. Na dost ka wait, jab mann kare tab practice.

> **Ye workflow kya karta hai**
>
> - Tum apne Telegram bot ko ek topic likhte ho, jaise HR ya Python (Telegram Trigger).
> - Basic LLM Chain topic dekh kar ek interview sawaal poochta hai; uske neeche Gemini model sochta hai.
> - Tum us sawaal ke message pe Reply karke jawab bhejte ho. Telegram reply ke saath purana sawaal bhi bhejta hai, to chain jawab ko us sawaal se jod leta hai.
> - Chain ek achhi cheez aur ek sudhaar ka tip likhta hai; Telegram Send Message wahi chat mein bhej deta hai.

![n8n ke official preview mein poora workflow: New Message > Interview Coach > Gemini Model > Send Reply](https://bloopcommits.pages.dev/coach/workflow.png)

_Poora workflow, n8n ke official preview mein (workflow hamara, screen n8n ka)._

Reel mein jo 'HR', 'Apne baare mein batao.' aur tip dikhe, wo banaya hua example hai. Ye starter workflow hai, n8n ke official preview mein dikhaya gaya. Credentials khud jodne padte hain, aur chalu karne se pehle ek test message se zaroor check karo.

## Node by node (n8n ke official docs ke hisaab se)

| Node (canvas par naam) | n8n node | Kya karta hai | Tumhe kya set karna hai |
| --- | --- | --- | --- |
| New Message | Telegram Trigger | Telegram ke events par workflow shuru karta hai, jaise tumhara Message. | Telegram credential (BotFather ka token). Event: Message. |
| Interview Coach | Basic LLM Chain | Prompt set karta hai jo model use karega. Prompt 'Define below' mein text ya expression de sakte ho. | Prompt (neeche diya hai). |
| Gemini Model | Google Gemini Chat Model (sub-node) | Google ke Gemini chat models ko chain mein use karta hai. Dropdown mein wahi models dikhte hain jo tumhare account ko available hain. | Google Gemini(PaLM) credential: API key Google AI Studio se. Model dropdown se chuno. |
| Send Reply | Telegram, Send Message | Chat mein message bhejta hai. Text ki limit 4096 characters. | Chat ID aur Text. 'Append n8n Attribution' default on hota hai; workflow mein off rakha hai. |

## Trick: memory node ke bina

Telegram Bot API ke Message object mein reply_to_message field hota hai: 'For replies in the same chat and message thread, the original message.' Yani jab tum bot ke sawaal pe Reply karte ho, bot ko tumhara jawab aur saath mein wo original sawaal dono milte hain. Is workflow ka prompt isi field se sawaal padhta hai (message.reply_to_message.text). Isliye is starter mein alag memory node nahi lagaya. (Ye hamare workflow ka design hai, n8n ka koi rule nahi.)

> **Banaya hua example**
>
> - Tum likhte ho:  HR
> - Bot poochta hai:  Apne baare mein batao.
> - Tum us message pe Reply karke likhte ho:  Main fresher hoon, coding seekhi hai.
> - Bot ka tip:  Achha intro. Sudhaar: ek project ka example jodo.

## Kya chahiye

- n8n: n8n Cloud (n8n hosting sambhalta hai) ya apna self-host (infrastructure tum khud provide aur manage karte ho). Plans aur kharche ke liye n8n ka official pricing page dekho.
- Telegram bot ka Access Token: Telegram par BotFather se chat shuru karo, /newbot likho, naam aur username do (username 'bot' par khatam hona chahiye, 5 se 32 characters), aur jo token mile use n8n mein Access Token ki jagah daalo.
- Gemini API key: Google AI Studio mein API Key page par Create API Key dabao, aur key n8n credential mein daalo.
- n8n ka HTTPS URL: Telegram Trigger ke liye n8n docs likhte hain ki webhook ke liye HTTPS URL dena zaroori hai. Self-host mein WEBHOOK_URL environment variable mein apne n8n ka public URL daalna padta hai.

## Step by step

1. n8n kholo (Cloud ya apna self-host) aur ek naya workflow banao.
2. Neeche wala starter workflow Import from File se laao, ya JSON copy karke canvas par paste karo.
3. Telegram bot banao (BotFather, /newbot) aur token copy karo. Telegram Trigger aur Send Reply dono nodes mein Telegram credential chuno ya naya banao: token Access Token mein jaata hai.
4. Google AI Studio mein Gemini API key banao. Gemini Model node mein Google Gemini(PaLM) credential banao aur key daalo. Model dropdown se ek model chuno.
5. n8n ka HTTPS URL set karo (self-host mein WEBHOOK_URL). Telegram ek app par sirf ek webhook allow karta hai: test URL aur production URL ke beech badloge to purana webhook overwrite hota hai.
6. Interview Coach node mein prompt dekho. Chaho to 'freshers' ki jagah apna role likho.
7. Bot ko ek topic bhejo, jaise HR. Bot ka sawaal aana chahiye.
8. Bot ke SAWAAL wale message pe Reply karo aur apna jawab likho. Tip aana chahiye. Agar reply khali aaye, to Send Reply ke Text mein Expression editor se Interview Coach ka output wala field chuno.
9. Sab sahi chale, tab workflow ko active karo.

[Workflow file download karo](https://bloopcommits.pages.dev/coach/bloopcommits-coach-workflow.json) - .json file · n8n mein Import from File · download na ho to neeche Copy dabao

**starter workflow: n8n mein Import from File / paste karo, credentials khud jodo, ek message se test karo**

```
{
  "name": "Interview Coach Bot Starter",
  "nodes": [
    {"id":"1","name":"New Message","type":"n8n-nodes-base.telegramTrigger","typeVersion":1.2,"position":[0,0],"parameters":{"updates":["message"],"additionalFields":{}}},
    {"id":"2","name":"Interview Coach","type":"@n8n/n8n-nodes-langchain.chainLlm","typeVersion":1.6,"position":[320,0],"parameters":{"promptType":"define","text":"=You are a friendly interview coach for freshers. Reply in simple Hinglish, at most 4 short lines.\n\nIf \"Question asked\" below is empty, the message is a TOPIC: ask ONE interview question on that topic and nothing else.\nIf \"Question asked\" has text, the message is the candidate's ANSWER to it: say one thing that was good and ONE thing to improve. Never promise a job or a result.\n\nQuestion asked: {{ $json.message.reply_to_message ? $json.message.reply_to_message.text : '' }}\nMessage: {{ $json.message.text }}"}},
    {"id":"3","name":"Gemini Model","type":"@n8n/n8n-nodes-langchain.lmChatGoogleGemini","typeVersion":1,"position":[340,260],"parameters":{"options":{}}},
    {"id":"4","name":"Send Reply","type":"n8n-nodes-base.telegram","typeVersion":1.2,"position":[700,0],"parameters":{"chatId":"={{ $('New Message').item.json.message.chat.id }}","text":"={{ $json.text }}","additionalFields":{"appendAttribution":false}}}
  ],
  "connections": {"New Message":{"main":[[{"node":"Interview Coach","type":"main","index":0}]]},"Interview Coach":{"main":[[{"node":"Send Reply","type":"main","index":0}]]},"Gemini Model":{"ai_languageModel":[[{"node":"Interview Coach","type":"ai_languageModel","index":0}]]}}
}
```

Is JSON mein credentials nahi hain: Telegram aur Gemini credential tumhe khud chunne hain, aur Gemini model dropdown se ek model bhi tumhe chunna hai.

## System prompt (Interview Coach node ka text)

**Basic LLM Chain ka prompt: topic aaye to sawaal, jawab aaye to tip**

```
You are a friendly interview coach for freshers. Reply in simple Hinglish, at most 4 short lines.

If "Question asked" below is empty, the message is a TOPIC: ask ONE interview question on that topic and nothing else.
If "Question asked" has text, the message is the candidate's ANSWER to it: say one thing that was good and ONE thing to improve. Never promise a job or a result.

Question asked: {{ $json.message.reply_to_message ? $json.message.reply_to_message.text : '' }}
Message: {{ $json.message.text }}
```

Prompt 'Question asked' khali hone par message ko topic maanta hai, aur bhara hone par jawab. Isme likha hai ki naukri ya result ka vaada kabhi na kare. Phir bhi AI galat ya zyada tareef wala feedback de sakta hai: use practice ki tarah lo.

## Topic list (suggestions)

| Topic likho | Kis liye |
| --- | --- |
| HR | Intro, strengths, weaknesses, "hum aapko kyun lein" jaise common sawaal |
| Python | Basics: list, dictionary, loop, function |
| SQL | Select, join, group by jaisi basics |
| Excel | Formulas, lookup, pivot table ki basics |
| Project | Apne project ko 1 minute mein samjhana |
| Communication | Clear aur chhota jawab dena |

Ye topics sirf suggestion hain (banaya hua list). Koi bhi topic likh ke dekh sakte ho; bot ka sawaal model ki soch pe hai, har baar alag ho sakta hai.

## Practice plan (suggestion, guarantee nahi)

1. Ek topic chuno aur bot se ek sawaal lo.
2. Pehle jawab apne shabdon mein bolke socho, phir type karke Reply karo.
3. Tip padho aur sirf EK cheez sudharo.
4. Jawab dobara likho, is baar wo ek sudhaar ke saath.
5. Doosre topic par jao. Har topic ke baad ek line apni notebook mein likho: kya sudhara.

## Aam galtiyan

- Jawab ko Reply ki jagah normal message bhej dena. Tab bot ko purana sawaal nahi milta aur wo use naya topic maan leta hai. Sawaal wale message pe Reply karo.
- Tip wale message pe Reply kar dena: prompt use bhi jawab maan lega. Sirf sawaal pe Reply karo.
- Telegram Trigger par HTTPS URL nahi hai, ya self-host mein WEBHOOK_URL set nahi hai. Telegram ke liye URL HTTPS hi hona chahiye.
- Test aur production URL ke beech badalna. Telegram ek app par sirf ek webhook allow karta hai, to purana webhook overwrite ho jaata hai.
- Telegram bot ka token galat hai. Token BotFather se aata hai. Send Reply mein Chat ID bhi dekho.
- Gemini Model node mein credential ya model chuna hi nahi.
- Test kiye bina workflow chalu kar dena. Pehle ek topic aur ek reply se check karo.

> **Privacy aur safety**
>
> - AI ka feedback practice hai: wo galat ho sakta hai, aur wo naukri nahi dilata. Ye advice hai, guarantee nahi.
> - Unpaid Gemini API par Google ke terms ke hisaab se tumhara bheja hua content aur generated responses Google ke products ko provide, improve aur develop karne mein use ho sakte hain, aur human reviewers input / output padh sakte hain. Google ka kehna hai: sensitive, confidential ya personal information Unpaid Services ko mat bhejo. Isliye apne jawab mein personal data (phone, address, ID) ya company ke secret mat likho.
> - Accounts aur credentials tumhare apne hain: bot token aur Gemini key kisi ko share mat karo.
> - Ye starter workflow hai. Ek test message se check karo, phir chalu karo.

## n8n kahan chalayein (plans badalte rehte hain, official page dekho)

| Option | n8n ke official pages ke hisaab se |
| --- | --- |
| n8n Cloud | Fully managed: n8n hosting, updates aur scaling sambhalta hai, install nahi karna padta. |
| Self-host | Setup (npm, Docker ya server) tumhe karna hai aur infrastructure tum khud provide aur manage karte ho. |

Dono mein Telegram Trigger ko HTTPS URL chahiye. Plans aur kharche n8n ke official pricing page par dekho; yahan koi price nahi likha.

> **Yaad rakho**
>
> - Telegram Trigger -> Basic LLM Chain (Gemini model) -> Telegram Send Message.
> - Topic bhejo, sawaal pe Reply karke jawab bhejo, ek tip lo, ek cheez sudharo.
> - AI ka feedback practice hai: galat ho sakta hai, naukri ki guarantee nahi.
> - Unpaid Gemini API par personal ya confidential data nahi.
> - Accounts aur credentials khud jodo; n8n mein HTTPS URL chahiye.
> - Starter workflow hai: ek message se test karo.

## Official links

- [n8n docs: Telegram Trigger](https://docs.n8n.io/integrations/builtin/trigger-nodes/n8n-nodes-base.telegramtrigger/)
- [n8n docs: Telegram Trigger common issues (HTTPS, WEBHOOK_URL)](https://docs.n8n.io/integrations/builtin/trigger-nodes/n8n-nodes-base.telegramtrigger/common-issues/)
- [n8n docs: Basic LLM Chain](https://docs.n8n.io/integrations/builtin/cluster-nodes/root-nodes/n8n-nodes-langchain.chainllm/)
- [n8n docs: Google Gemini Chat Model](https://docs.n8n.io/integrations/builtin/cluster-nodes/sub-nodes/n8n-nodes-langchain.lmchatgooglegemini/)
- [n8n docs: Telegram node, Send Message](https://docs.n8n.io/integrations/builtin/app-nodes/n8n-nodes-base.telegram/message-operations/)
- [n8n docs: Telegram credentials (BotFather token)](https://docs.n8n.io/integrations/builtin/credentials/telegram/)
- [n8n docs: Google Gemini(PaLM) credentials (API key)](https://docs.n8n.io/integrations/builtin/credentials/googleai/)
- [n8n docs: n8n Cloud ya self-host](https://docs.n8n.io/choose-how-to-use-n8n/)
- [Telegram Bot API: Message object (reply_to_message)](https://core.telegram.org/bots/api#message)
- [Google AI Studio: API key](https://aistudio.google.com/apikey)
- [Gemini API terms (Unpaid Services / Paid Services)](https://ai.google.dev/gemini-api/terms)

---
More free guides: https://bloopcommits.pages.dev/