# AI Agent MCP Server
### ChatGPT Agent Reports ko MongoDB mein store karo — Step by Step Guide

---

## Yeh Kya Hai?

ChatGPT ke scheduled agents kaam karte hain aur reports apni chat mein store karte hain.
Yeh server ek bridge hai jo:
- ChatGPT Agent se data receive karta hai (Custom MCP ya REST API)
- MongoDB Atlas mein permanently store karta hai
- Kisi bhi time data retrieve karne deta hai

```
⏰ ChatGPT Scheduled Agent
        ↓
🔧 Yeh MCP Server (/mcp endpoint)
        ↓
💾 MongoDB Atlas Database
        ↓
📊 Kabhi bhi data dekho (API ya Atlas Dashboard)
```

---

## STEP 1 — MongoDB Atlas Setup (Free)

1. **cloud.mongodb.com** pe jao
2. Free account banao
3. **New Project** → **Create Cluster** → **M0 Free** select karo
4. Username aur Password set karo (yaad rakhna!)
5. **Network Access** → **Add IP Address** → **Allow from anywhere** (0.0.0.0/0)
6. **Connect** → **Drivers** → Node.js → Connection string copy karo:
   ```
   mongodb+srv://USERNAME:PASSWORD@cluster0.xxxxx.mongodb.net/ai_agents
   ```
7. Yeh string save kar lo — baad mein chahiye hogi

---

## STEP 2 — GitHub pe Upload Karo

```bash
# Project folder mein jao
cd ai-agent-mcp

# Git initialize karo
git init
git add .
git commit -m "Initial commit"

# GitHub pe new repository banao: github.com/new
# Phir yeh commands chalao:
git remote add origin https://github.com/TERA_USERNAME/ai-agent-mcp.git
git push -u origin main
```

---

## STEP 3 — Railway pe Deploy Karo (Free)

1. **railway.app** pe jao → Free account banao
2. **New Project** → **Deploy from GitHub repo**
3. Apna `ai-agent-mcp` repo select karo
4. **Variables** tab mein yeh add karo:
   ```
   MONGO_URI = mongodb+srv://USERNAME:PASSWORD@cluster0.xxxxx.mongodb.net/ai_agents
   PORT = 3000
   ```
5. **Deploy** click karo
6. Kuch minutes mein URL milega jaise:
   ```
   https://ai-agent-mcp-production.up.railway.app
   ```
7. Browser mein kholo → `{"status": "✅ AI Agent MCP Server is running!"}` dikhega

**Yeh URL save kar lo — ChatGPT mein daalna hai!**

---

## STEP 4 — ChatGPT mein Custom MCP Connect Karo

1. **chatgpt.com** → Settings → **Developer Mode ON** karo
2. Apna Agent open karo (Edit)
3. **Apps** → **Custom MCP** → Enable
4. MCP Server URL daalo:
   ```
   https://ai-agent-mcp-production.up.railway.app/mcp
   ```
5. Save karo → Tools appear honge:
   - `save_data`
   - `get_data`
   - `get_latest`
   - `log_activity`

---

## STEP 5 — Agent Instructions Update Karo

Agent ke **Instructions** mein yeh add karo:

```
IMPORTANT: Har task complete karne ke baad HAMESHA yeh karo:

1. Apna kaam karo (SEO check / analysis / report)
2. save_data tool call karo:
   - agentName: "[TERA AGENT KA NAAM]"
   - taskType: "[kya kiya, e.g. seo_scan]"
   - status: "success" ya "failed"
   - payload: {
       summary: "kya mila",
       details: [...findings...],
       recommendations: [...suggestions...]
     }
   - metadata: {
       url: "[website jo check ki]",
       model: "gpt-4",
       duration: "[kitna time laga]"
     }

3. Kabhi bhi sirf chat mein result mat rakho
4. Hamesha database mein save karo
```

---

## STEP 6 — Data Dekho

### Option A: MongoDB Atlas Dashboard
- cloud.mongodb.com → Apna cluster → Browse Collections
- `ai_agents` database → `agentdatas` collection

### Option B: API se
```bash
# Sab agents dekho
GET https://tera-server.up.railway.app/api/agents

# Specific agent ki reports
GET https://tera-server.up.railway.app/api/reports/SEO%20Agent

# Latest report
GET https://tera-server.up.railway.app/api/latest/SEO%20Agent

# Filter karo
GET https://tera-server.up.railway.app/api/reports/SEO%20Agent?taskType=seo_scan&limit=5
```

---

## API Reference

### POST /api/save
```json
{
  "agentName": "SEO Agent",
  "taskType": "seo_scan",
  "status": "success",
  "payload": {
    "website": "example.com",
    "score": 85,
    "issues": ["Missing meta description", "Slow page speed"],
    "recommendations": ["Add meta tags", "Optimize images"]
  },
  "metadata": {
    "url": "https://example.com",
    "checkedAt": "2024-01-15T09:00:00Z"
  }
}
```

### GET /api/reports/:agentName
Query params: `limit`, `page`, `taskType`, `status`

### GET /api/latest/:agentName

### GET /api/agents

---

## Local Testing (Optional)

```bash
# Dependencies install karo
npm install

# .env file banao
cp .env.example .env
# .env mein MONGO_URI daalo

# Server start karo
npm run dev

# Test karo
curl -X POST http://localhost:3000/api/save \
  -H "Content-Type: application/json" \
  -d '{"agentName":"Test Agent","taskType":"test","payload":{"message":"Hello!"}}'
```

---

## Project Structure

```
ai-agent-mcp/
├── server.js          ← Main entry point
├── package.json       ← Dependencies
├── railway.toml       ← Railway deploy config
├── .env.example       ← Environment variables template
├── .gitignore
├── models/
│   └── AgentData.js   ← MongoDB schema
├── routes/
│   └── api.js         ← REST API endpoints
└── mcp/
    └── tools.js       ← MCP tools (save_data, get_data, etc.)
```

---

## Problem Aaye Toh?

| Problem | Solution |
|---------|----------|
| MongoDB connect nahi | IP whitelist check karo (0.0.0.0/0 hona chahiye) |
| Railway deploy fail | Logs check karo → Variables mein MONGO_URI sahi daala? |
| ChatGPT MCP nahi dikha | Developer Mode ON hai? Business/Plus plan chahiye |
| Tools appear nahi | MCP URL mein `/mcp` path daala? |