Skip to main content
Follow these best practices to get the most out of Screen Answerer while managing your API quota efficiently and maintaining security.

​
API quota management

Managing your Google Gemini API quota is crucial for uninterrupted service, especially on the free tier.

​
Understanding quotas

The Gemini API free tier includes:
  • 15 requests per minute (RPM)
  • 1,500 requests per day (RPD)
  • 4 million tokens per day
Screen Answerer’s 5-second monitoring interval means you’ll make up to 12 requests per minute when actively monitoring. This stays safely under the 15 RPM limit.

​
Quota optimization strategies

The most effective way to preserve quota:
  • Click Stop Monitoring between quizzes
  • Don’t leave monitoring running overnight or during breaks
  • Only start monitoring right before you expect questions
Impact: Saves ~720 requests per hour when monitoring is stopped.
Capture only the window containing questions:
  • Reduces image size sent to API
  • Faster processing times
  • Better detection accuracy
  • Lower bandwidth usage
Impact: ~30-50% reduction in API processing time and token usage.
Use the appropriate model for your needs:
  • gemini-2.0-flash-lite - Best for simple multiple-choice questions, uses fewer tokens
  • gemini-2.0-flash - Better for complex questions, explanations, or calculations
Impact: Lite model uses approximately 40-60% fewer tokens than flash.
Start with flash-lite. Only upgrade to flash if you notice accuracy issues with complex questions.
Track your API consumption:
  1. Visit Google Cloud Console
  2. Navigate to APIs & Services → Dashboard
  3. Select Generative Language API
  4. View quota usage graphs
Set up quota alerts to notify you before hitting limits.

​
Model selection guidelines

Choosing the right model balances speed, accuracy, and cost.

​
When to use gemini-2.0-flash-lite

Best for:
  • Multiple-choice questions
  • Simple fact-based questions
  • True/false questions
  • Quick lookups
  • High-frequency monitoring
Advantages:
  • Faster response times (200-400ms average)
  • Lower token consumption
  • Higher quota efficiency
  • Suitable for most quiz scenarios
Example use case: Monitoring a multiple-choice quiz where you need quick answers to straightforward questions.

​
When to use gemini-2.0-flash

Best for:
  • Complex problem-solving
  • Mathematical calculations
  • Essay-style questions
  • Questions requiring detailed explanations
  • Technical or specialized topics
Advantages:
  • Higher accuracy on complex questions
  • Better reasoning capabilities
  • More detailed responses
  • Superior context understanding
Example use case: Technical certification exams with scenario-based questions requiring analysis.
You can switch models anytime in Settings → Model tab. The change takes effect immediately for the next request.

​
Security best practices

Protect your API key and maintain privacy while using Screen Answerer.

​
API key security

1

Never share your API key

Treat your API key like a password:
  • Don’t post it in forums or chat
  • Don’t commit it to version control
  • Don’t share it with others
  • Don’t include it in screenshots
2

Use API restrictions

Add security restrictions in Google Cloud Console:
  1. Go to APIs & Services → Credentials
  2. Click on your API key
  3. Under Application restrictions, select:
    • HTTP referrers and add your domain, OR
    • IP addresses and add your network IP
  4. Under API restrictions, select Restrict key and choose only “Generative Language API”
Application restrictions may interfere with Screen Answerer if configured incorrectly. Test thoroughly after applying restrictions.
3

Rotate keys regularly

Best practice for long-term security:
  • Regenerate your API key every 3-6 months
  • Delete old keys after rotating
  • Update Screen Answerer immediately with new key
4

Monitor for unauthorized usage

Check your API usage regularly:
  • Look for unexpected spikes
  • Verify request patterns match your usage
  • Set up billing alerts if using paid tier
  • Regenerate key immediately if compromise is suspected

​
Screen capture privacy

Screen monitoring captures everything in the selected area. Be careful about sensitive information.
Privacy checklist:
  • ✅ Only capture the window with quiz content
  • ✅ Close sensitive documents before monitoring
  • ✅ Disable notifications that might appear on screen
  • ✅ Use a dedicated browser profile for quizzes
  • ❌ Don’t capture your entire screen unnecessarily
  • ❌ Don’t monitor when handling confidential information
  • ❌ Don’t share your screen capture preview publicly

​
Data handling

Understand how your data flows:
  1. Screenshot capture - Taken locally in your browser
  2. Transmission - Sent directly to Google Gemini API (not stored by Screen Answerer)
  3. Processing - Analyzed by Google’s servers
  4. Response - Answer sent back to your browser
  5. Cleanup - Screenshot immediately deleted from temporary storage (server.js:229, 269, 310, 341)
Screen Answerer server acts as a proxy but does not log or store your images. All screenshots are deleted after processing using history.safelyDeleteFile() and history.markFileProcessed() functions.

​
Performance optimization

Improve response times and reduce resource usage.

​
Client-side optimizations

Chrome and Firefox offer the best performance:
  • Native screen capture APIs
  • Optimized image encoding
  • Better WebSocket support
  • Lower CPU overhead
Avoid older browsers or those with limited screen capture support.
Free up system resources:
  • Close unused browser tabs
  • Quit background applications
  • Disable resource-intensive extensions
  • Ensure adequate RAM availability
Minimum recommended: 4GB RAM, modern CPU
Select efficient capture options:
  • Window capture over entire screen
  • Capture at native resolution (don’t upscale)
  • Ensure good lighting for better image quality
  • Position questions centrally in the window

​
Network optimization

  • Stable connection - Use wired Ethernet if possible
  • Bandwidth - Minimum 5 Mbps recommended for smooth operation
  • Latency - Lower ping to Google servers improves response times
  • Firewall - Ensure generativelanguage.googleapis.com is not blocked

​
Server-side considerations

Screen Answerer includes several performance features:
  • Rate limiting - 5-second minimum between requests (server.js:87-102)
  • Retry logic - Automatic retry with exponential backoff (server.js:135-170)
  • Quota tracking - Prevents hitting API limits (server.js:104-122)
  • Image cleanup - Immediate deletion of temporary files (server.js:227-230)
These features are automatic. You don’t need to configure anything, but understanding them helps troubleshoot issues.

​
Error handling

Handle common errors gracefully and know when to retry.

​
Rate limit errors

Error: “Rate limit exceeded” Solutions:
  1. Wait 5-10 seconds before next action
  2. Reduce monitoring frequency (built-in)
  3. Check if you’re running multiple instances
  4. Verify you haven’t exceeded daily quota
Screen Answerer automatically implements retry logic with exponential backoff (server.js:135-170), so most rate limit errors resolve themselves.

​
API key errors

Error: “Invalid API key” or “API key error” Solutions:
  1. Verify key format (starts with AIza, 39 characters total)
  2. Check for extra spaces when copying
  3. Ensure API is enabled in Google Cloud
  4. Try regenerating the key
  5. Confirm billing is set up (even for free tier)

​
Detection issues

Problem: Questions not being detected Solutions:
  1. Ensure question text is clear and readable
  2. Wait at least 5 seconds after question appears
  3. Try capturing a larger window area
  4. Switch to gemini-2.0-flash for better detection
  5. Check image preview to verify capture quality

​
Testing and validation

Verify your setup before critical use.

​
API key validation

Test your API key: 
curl -X POST \
  "https://generativelanguage.googleapis.com/v1beta/models/gemini-2.0-flash-lite:generateContent?key=YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"contents":[{"parts":[{"text":"What is 2+2?"}]}]}'
Expected response should include "text": "4" or similar.

​
Monitor testing

  1. Start monitoring
  2. Display a sample quiz question on screen
  3. Wait 5-10 seconds
  4. Verify answer appears in ANSWER section
  5. Check Questions Processed counter increments

​
Performance baseline

Expected metrics:
  • Detection time: 5-8 seconds from question display
  • API response: 200-600ms (flash-lite) or 400-1000ms (flash)
  • CPU usage: Less than 10% on modern hardware
  • Memory: Less than 100MB additional RAM
If metrics significantly exceed these ranges, review the performance optimization section above.

​
Compliance and ethical use

​
Academic integrity

Using Screen Answerer to cheat on exams or assignments violates academic integrity policies at most institutions. This tool is for educational assistance and study purposes only.
Appropriate uses:
  • Self-study and practice quizzes
  • Tutoring and educational assistance
  • Reviewing quiz answers after completion
  • Learning tool for understanding concepts
Inappropriate uses:
  • Cheating on graded exams
  • Academic dishonesty
  • Violating testing policies
  • Circumventing proctored exam systems

​
API terms of service

Ensure your use complies with:

​
Next steps

API key setup

Get your Google Gemini API key

Screen monitoring

Start using the screen monitor

Model selection

Choose the right Gemini model

Troubleshooting

Solve common issues

Build docs developers (and LLMs) love

Get started for freeTalk to us