Re_Backend/API_KEY_TROUBLESHOOTING.md

# 🔑 API Key Troubleshooting Guide

## ⚠️ Problem: Getting 404 Errors for ALL Claude Models

If you're getting 404 "model not found" errors for **multiple Claude models**, the issue is likely with your Anthropic API key, not the model versions.

---

## 🔍 Step 1: Verify Your API Key

### Check Your API Key Status

1. Go to: https://console.anthropic.com/
2. Log in to your account
3. Navigate to **Settings** → **API Keys**
4. Check:
   - ✅ Is your API key active?
   - ✅ Does it have an active billing method?
   - ✅ Have you verified your email?
   - ✅ Are there any usage limits or restrictions?

### API Key Tiers

Anthropic has different API access tiers:

| Tier | Access Level | Requirements |
|------|-------------|--------------|
| **Free Trial** | Limited models, low usage | Email verification |
| **Paid Tier 1** | All Claude 3 models | Add payment method, some usage |
| **Paid Tier 2+** | All models + higher limits | More usage history |

**If you just created your API key:**
- You might need to add a payment method
- You might need to make a small payment first
- Some models might not be available immediately

---

## 🎯 Step 2: Try the Most Basic Model (Claude 3 Haiku)

I've changed the default to **`claude-3-haiku-20240307`** - this should work with ANY valid API key.

### Restart Your Backend

**IMPORTANT:** You must restart the server for changes to take effect.

```bash
# Stop the current server (Ctrl+C)
# Then start again:
cd Re_Backend
npm run dev
```

### Check the Startup Logs

Look for this line:
```
[AI Service] ✅ Claude provider initialized with model: claude-3-haiku-20240307
```

### Test Again

Try generating a conclusion. You should see in logs:
```
[AI Service] Generating with Claude model: claude-3-haiku-20240307
```

---

## 🔧 Step 3: Check for Environment Variable Overrides

Your `.env` file might be overriding the default model.

### Check Your `.env` File

Open `Re_Backend/.env` and look for:
```bash
CLAUDE_MODEL=...
```

**If it exists:**
1. **Delete or comment it out** (add `#` at the start)
2. **Or change it to Haiku:**
   ```bash
   CLAUDE_MODEL=claude-3-haiku-20240307
   ```
3. **Restart the server**

---

## 🐛 Step 4: Verify API Key is Loaded

Add this temporary check to see if your API key is being loaded:

### Option A: Check Logs on Startup

When you start the server, you should see:
```
[AI Service] ✅ Claude provider initialized with model: claude-3-haiku-20240307
```

If you DON'T see this:
- Your API key might be missing or invalid
- Check `.env` file has: `CLAUDE_API_KEY=sk-ant-api03-...`

### Option B: Test API Key Manually

Create a test file `Re_Backend/test-api-key.js`:

```javascript
const Anthropic = require('@anthropic-ai/sdk');
require('dotenv').config();

const apiKey = process.env.CLAUDE_API_KEY || process.env.ANTHROPIC_API_KEY;

console.log('API Key found:', apiKey ? 'YES' : 'NO');
console.log('API Key starts with:', apiKey ? apiKey.substring(0, 20) + '...' : 'N/A');

async function testKey() {
  try {
    const client = new Anthropic({ apiKey });
    
    // Try the most basic model
    const response = await client.messages.create({
      model: 'claude-3-haiku-20240307',
      max_tokens: 100,
      messages: [{ role: 'user', content: 'Say hello' }]
    });
    
    console.log('✅ API Key works!');
    console.log('Response:', response.content[0].text);
  } catch (error) {
    console.error('❌ API Key test failed:', error.message);
    console.error('Error details:', error);
  }
}

testKey();
```

Run it:
```bash
cd Re_Backend
node test-api-key.js
```

---

## 💡 Step 5: Alternative - Use OpenAI or Gemini

If your Anthropic API key has issues, you can switch to another provider:

### Option A: Use OpenAI

1. **Get OpenAI API key** from: https://platform.openai.com/api-keys

2. **Add to `.env`:**
   ```bash
   AI_PROVIDER=openai
   OPENAI_API_KEY=sk-...
   ```

3. **Install OpenAI SDK:**
   ```bash
   cd Re_Backend
   npm install openai
   ```

4. **Restart server**

### Option B: Use Google Gemini

1. **Get Gemini API key** from: https://makersuite.google.com/app/apikey

2. **Add to `.env`:**
   ```bash
   AI_PROVIDER=gemini
   GEMINI_API_KEY=...
   ```

3. **Install Gemini SDK:**
   ```bash
   cd Re_Backend
   npm install @google/generative-ai
   ```

4. **Restart server**

---

## 🎯 Quick Checklist

- [ ] My Anthropic API key is valid and active
- [ ] I have a payment method added (if required)
- [ ] My email is verified
- [ ] I've deleted/commented out `CLAUDE_MODEL` from `.env` (or set it to haiku)
- [ ] I've **restarted the backend server completely**
- [ ] I see the correct model in startup logs
- [ ] I've tested with the test script above

---

## 🆘 Still Not Working?

### Check Your API Key Format

Valid format: `sk-ant-api03-XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX`

- Must start with `sk-ant-`
- Must be quite long (80+ characters)
- No spaces or line breaks

### Get a New API Key

1. Go to https://console.anthropic.com/settings/keys
2. Delete old key
3. Create new key
4. Add payment method if prompted
5. Update `.env` with new key
6. Restart server

### Contact Anthropic Support

If nothing works:
- Email: support@anthropic.com
- Check: https://status.anthropic.com/ (for service issues)
- Community: https://anthropic.com/community

---

## 🎯 Current System Default

The system now defaults to:
```
claude-3-haiku-20240307
```

This is the **most basic Claude model** that should work with **any valid API key**, even free tier.

If even Haiku doesn't work, there's a fundamental issue with your Anthropic API key or account status.

---

## ✅ Success Indicators

When everything is working correctly, you should see:

1. **On server startup:**
   ```
   [AI Service] ✅ Claude provider initialized with model: claude-3-haiku-20240307
   ```

2. **When generating conclusion:**
   ```
   [AI Service] Generating with Claude model: claude-3-haiku-20240307
   ```

3. **In response:**
   ```
   [AI Service] ✅ Conclusion generated successfully
   ```

No 404 errors! ✅