Skip to main content
ScryCLI requires authentication to access AI models through OpenRouter. This guide walks you through setting up your API credentials and understanding the authentication system.

​
Authentication Overview

ScryCLI uses a two-layer authentication system:
  1. Model Authentication: OpenRouter API key for AI model access
  2. User Authentication: JWT-based token system (optional)
All authentication data is stored locally in ~/.scrycli/config.json.

​
Quick Start

1

Get OpenRouter API Key

Sign up at OpenRouter and generate an API key:
  1. Create an account at https://openrouter.ai/
  2. Navigate to API Keys section
  3. Generate a new API key
  4. Copy the key (starts with sk-or-)
2

Launch ScryCLI

Run the CLI tool:
scrycli
If no model is configured, you’ll be prompted to set one up.
3

Configure Model

When prompted, enter your credentials:
Select AI Model: anthropic/claude-3-opus
Enter API Key: sk-or-v1-your-api-key-here
This information is saved to ~/.scrycli/config.json:
{
  "model": {
    "modelName": "anthropic/claude-3-opus",
    "modelKey": "sk-or-v1-your-api-key-here"
  }
}
4

Verify Setup

Test your configuration:
> Analyze a simple JavaScript file
If authentication is successful, the AI will respond.

​
Configuration File

ScryCLI stores all configuration in ~/.scrycli/config.json: 
// From src/config/configManage.ts
const configPath = path.join(os.homedir(), ".scrycli", "config.json");

if(!fs.existsSync(configPath)){
  fs.mkdirSync(path.dirname(configPath), {recursive:true})
  fs.writeFileSync(configPath, JSON.stringify({}));
}

​
Config Structure

{
  "model": {
    "modelName": "anthropic/claude-3-opus",
    "modelKey": "sk-or-v1-..."
  },
  "user": {
    "token": "eyJhbGciOiJIUzI1NiIs..."
  }
}

​
Config Fields

  • model.modelName: The AI model identifier (e.g., anthropic/claude-3-opus, openai/gpt-4)
  • model.modelKey: Your OpenRouter API key
  • user.token: JWT token for user authentication (optional)

​
Supported AI Models

ScryCLI works with any model available on OpenRouter: 
# High performance (expensive)
anthropic/claude-3-opus
anthropic/claude-3-sonnet
openai/gpt-4-turbo

# Balanced (moderate cost)
anthropic/claude-3-haiku
openai/gpt-3.5-turbo
meta-llama/llama-3-70b

# Budget-friendly
meta-llama/llama-3-8b
mistralai/mistral-7b

​
Model Selection

Choose based on your needs: For Complex Code Analysis: 
{
  "model": {
    "modelName": "anthropic/claude-3-opus",
    "modelKey": "your-key"
  }
}
For Basic Operations: 
{
  "model": {
    "modelName": "anthropic/claude-3-haiku",
    "modelKey": "your-key"
  }
}
Start with Claude 3 Haiku for testing. It’s fast, affordable, and handles most ScryCLI tasks well. Upgrade to Opus for complex refactoring or architecture decisions.

​
Model Validation

ScryCLI validates your model configuration before making API calls: 
// From src/lib/isModelSelected.ts
const isModelSelected = (): boolean => {
  try{
    const config = getConfig();
    if (!config?.model?.modelName) return false;
    if (!config?.model?.modelKey) return false;
    return config.model.modelName !== '' && config.model.modelKey !== '';
  }catch(error){
    console.error(error);
    return false;
  }
};
This function checks:
  • Config file exists and is readable
  • model.modelName is set and non-empty
  • model.modelKey is set and non-empty

​
User Authentication (Optional)

ScryCLI includes an optional JWT-based user authentication system: 
// From src/lib/auth.ts
import jwt from 'jsonwebtoken';

const isAuthenticated = (): boolean => {
  try {
    const config = getConfig();

    if (!config?.user?.token) return false;
    if (!process.env.JWT_SECRET) throw new Error('JWT_SECRET is not defined');

    const decoded = jwt.verify(config.user.token, process.env.JWT_SECRET) as any;

    return decoded && typeof decoded === 'object' && 'userId' in decoded;
  } catch (error) {
    return false;
  }
};

​
Setting Up JWT Authentication

1

Set JWT Secret

Create a .env file in your project root:
JWT_SECRET=your-secure-secret-key-here
2

Generate User Token

Generate a JWT token for your user:
const jwt = require('jsonwebtoken');
const token = jwt.sign(
  { userId: 'your-user-id' },
  process.env.JWT_SECRET,
  { expiresIn: '30d' }
);
console.log(token);
3

Add to Config

Manually add the token to ~/.scrycli/config.json:
{
  "model": { /* ... */ },
  "user": {
    "token": "eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9..."
  }
}
User authentication is optional and not required for basic ScryCLI functionality. It’s primarily used for enterprise deployments or multi-user scenarios.

​
Managing Configuration

ScryCLI provides utilities for managing configuration: 
// From src/config/configManage.ts

// Get entire config
export const getConfig = () => {
  const config = JSON.parse(fs.readFileSync(configPath, "utf-8"));
  return config;
};

// Set a config value
export const setConfig = (key: string, value: any) => {
  const config = getConfig();
  config[key] = value;
  fs.writeFileSync(configPath, JSON.stringify(config, null, 2));
};

// Delete a config value
export const deleteConfig = (key: string) => {
  const config = getConfig();
  delete config[key];
  fs.writeFileSync(configPath, JSON.stringify(config, null, 2));
};

​
Manual Config Management

# View current config
cat ~/.scrycli/config.json

# Edit config
vim ~/.scrycli/config.json

# Reset config (delete and restart ScryCLI)
rm ~/.scrycli/config.json
scrycli

​
Changing Models

To switch AI models:

​
Method 1: Edit Config File

vim ~/.scrycli/config.json
Update the modelName: 
{
  "model": {
    "modelName": "anthropic/claude-3-haiku",
    "modelKey": "your-existing-key"
  }
}

​
Method 2: Reset and Reconfigure

rm ~/.scrycli/config.json
scrycli
# Follow setup prompts again

​
OpenRouter Integration

ScryCLI uses the OpenRouter SDK for AI model access: 
// From src/model/openRouter.ts
import { OpenRouter } from '@openrouter/sdk';
import { getConfig } from '../config/configManage.js';

const openRouterClient = new OpenRouter({
  apiKey: '',  // Set dynamically from config
});

export async function llmCall({
  prompt,
  systemPrompt,
}: llmCallParams): Promise<string> {
  const result = openRouterClient.callModel({
    model: `${getConfig().model.modelName}`,
    instructions: `${systemPrompt}`,
    input: `${prompt} \n\nFile Tree: ${fileTreeString}`,
  });
  const text = await result.getText();
  return text;
}

​
API Key Usage

The API key is read from config for each request: 
const config = getConfig();
const apiKey = config.model.modelKey;
Your API key is stored in plain text in ~/.scrycli/config.json. Ensure this file has appropriate permissions (0600 recommended).

​
Security Best Practices

​
1. Protect Your Config File

# Set restrictive permissions
chmod 600 ~/.scrycli/config.json

# Verify permissions
ls -la ~/.scrycli/config.json
# Should show: -rw-------

​
2. Use Environment Variables for JWT Secret

# .env (never commit this)
JWT_SECRET=your-very-secure-random-string-here

# .gitignore
echo ".env" >> .gitignore

​
3. Rotate API Keys Regularly

# Generate new key at OpenRouter
# Update config
vim ~/.scrycli/config.json
# Update modelKey value

​
4. Don’t Share Config Files

Never commit or share your ~/.scrycli/config.json file: 
# It contains sensitive credentials
{
  "model": {
    "modelKey": "sk-or-v1-..."  // Secret!
  }
}

​
Troubleshooting

​
Model Not Selected Error

Error: Model not configured
Solution: 
rm ~/.scrycli/config.json
scrycli
# Reconfigure model

​
Invalid API Key

Error: 401 Unauthorized
Solution:
  1. Verify your API key at OpenRouter
  2. Check for typos in config.json
  3. Ensure key hasn’t expired or been revoked

​
JWT Verification Failed

Error: JWT_SECRET is not defined
Solution: 
# Add to .env
echo "JWT_SECRET=your-secret" >> .env

# Or remove user token from config
vim ~/.scrycli/config.json
# Delete the "user" section

​
Config File Corrupted

Error: Unexpected token in JSON
Solution: 
# Backup if needed
cp ~/.scrycli/config.json ~/.scrycli/config.json.bak

# Reset
rm ~/.scrycli/config.json
scrycli

​
Environment Variables

ScryCLI uses dotenv for environment variables: 
// From src/lib/auth.ts
import dotenv from 'dotenv';
dotenv.config({ quiet: true });

​
Supported Variables

# .env
JWT_SECRET=your-jwt-secret-for-user-auth
The quiet: true option suppresses dotenv warnings if .env doesn’t exist.

​
Multiple Environments

For different projects or environments: 
# Development
export SCRYCLI_CONFIG=~/.scrycli/config.dev.json

# Production
export SCRYCLI_CONFIG=~/.scrycli/config.prod.json
Note: This requires modifying the source code to read from process.env.SCRYCLI_CONFIG.

​
Cost Management

Monitor your OpenRouter usage:
  1. Check Usage Dashboard: https://openrouter.ai/dashboard
  2. Set Budget Alerts: Configure in OpenRouter settings
  3. Choose Cost-Effective Models: Use Haiku for routine tasks, Opus for complex work

​
Cost Estimates (Approximate)

# Per 1M tokens (varies by model)
Claude 3 Opus:    $15-75
Claude 3 Sonnet:  $3-15
Claude 3 Haiku:   $0.25-1.25
GPT-4 Turbo:      $10-30
GPT-3.5 Turbo:    $0.50-1.50
For typical ScryCLI usage (file operations, small code analysis), Claude 3 Haiku provides the best value. Reserve Opus for complex architectural decisions.

​
Next Steps

Code Analysis

Start analyzing code with your configured model

File Management

Use file operations through natural language