File Attachments

File attachments allow you to provide additional context to the AI by including files with your prompts.

Attachment Types

The SDK supports several attachment types:

Type	Description	Use Cases
image	Image files	Screenshots, diagrams, UI mockups
text	Text files	Source code, configs, logs
data	Data files	JSON, CSV, structured data

Basic Usage

const result = await thread.run('Describe this image', {
  attachments: [
    { path: './screenshot.png', type: 'image' }
  ]
});

Image Attachments

Attach images for visual analysis:

// Single image
const result = await thread.run('What does this UI show?', {
  attachments: [
    { path: './mockup.png', type: 'image' }
  ]
});

// Multiple images
const result = await thread.run('Compare these two designs', {
  attachments: [
    { path: './design-v1.png', type: 'image' },
    { path: './design-v2.png', type: 'image' }
  ]
});

Text Attachments

Attach text files for code review or analysis:

// Code review
const result = await thread.run('Review this code for security issues', {
  attachments: [
    { path: './src/auth.ts', type: 'text', description: 'Authentication module' }
  ]
});

// Multiple files with context
const result = await thread.run('How do these modules interact?', {
  attachments: [
    { path: './src/user.ts', type: 'text', description: 'User model' },
    { path: './src/auth.ts', type: 'text', description: 'Auth service' },
    { path: './src/api.ts', type: 'text', description: 'API routes' }
  ]
});

Data Attachments

Attach structured data files:

// JSON data
const result = await thread.run('Analyze this API response', {
  attachments: [
    { path: './response.json', type: 'data' }
  ]
});

// Configuration analysis
const result = await thread.run('Review our build configuration', {
  attachments: [
    { path: './tsconfig.json', type: 'data' },
    { path: './package.json', type: 'data' }
  ]
});

File Attachment Interface

interface FileAttachment {
  /** Path to the file */
  path: string;

  /** Type of file content */
  type: 'image' | 'text' | 'data';

  /** Optional description for context */
  description?: string;
}

Practical Examples

Screenshot Analysis

async function analyzeScreenshot(screenshotPath: string) {
  const droid = new Droid();
  const thread = droid.startThread();

  const result = await thread.run(
    'Analyze this screenshot and identify any UI issues or improvements',
    {
      attachments: [
        { path: screenshotPath, type: 'image' }
      ]
    }
  );

  return result.finalResponse;
}

Code Review Workflow

async function reviewChanges(files: string[]) {
  const droid = new Droid();
  const thread = droid.startThread();

  const attachments = files.map(file => ({
    path: file,
    type: 'text' as const,
    description: `File: ${file}`
  }));

  const result = await thread.run(
    'Review these files for bugs, security issues, and code quality',
    { attachments }
  );

  return result.finalResponse;
}

Error Log Analysis

async function analyzeErrors(logPath: string) {
  const droid = new Droid();

  const result = await droid.exec(
    'Analyze this log file and summarize the errors',
    {
      attachments: [
        { path: logPath, type: 'text', description: 'Application error log' }
      ]
    }
  );

  return result.finalResponse;
}

Best Practices

1. Use descriptions - Help the AI understand file purpose
2. Limit file size - Large files may be truncated
3. Choose correct type - Affects how content is processed
4. Group related files - Attach files that work together
5. Be specific in prompts - Reference attachments in your prompt