DeepSec AI Agent

DeepSec AI Agent is your intelligent security assistant that helps you analyze vulnerabilities, understand threats, and get actionable recommendations. Available across the platform, it provides context-aware assistance for your security operations.

Accessing DeepSec AI Agent

From the Navigation Menu

Click AI Agent in the left sidebar under Services to open a dedicated AI conversation interface.

From Vulnerability Dashboard

Access the AI agent directly from any vulnerability card or the main vulnerability dashboard to get context-specific assistance.

From AI-Context Page

Navigate to Vulnerability > AI-Context to start a new conversation or view your recent chat history.

Key Features

Recent Chats

View and manage your conversation history with DeepSec AI Agent.

Features:

• Search Conversations: Find specific chats using the search bar
• Chat History: Access all your previous conversations with timestamps
• Continue Conversations: Resume any previous chat to maintain context
• Delete Chats: Remove conversations you no longer need

Chat Information:

• Chat title or first message preview
• Date and timestamp (e.g., 10/15/2025)
• AI model used (e.g., Deepseek Chat)
• Delete option for each conversation

AI Model Selection

Choose from multiple AI models based on your specific needs:

OpenAI: GPT-4o

Advanced model for reasoning and multimodal input. Best for complex security analysis requiring deep reasoning.

Deepseek Chat

Fast, general-purpose chat with strong reasoning. Recommended as the default model for most security queries.

Gemini 2.0-flash

Best for fast responses and quick summaries. Ideal when you need rapid answers to straightforward questions.

Cloudflare: Llama 70b

Accurate, open-source model for detailed answers. Excellent for comprehensive security analysis.

Cloudflare: Deepseek 32b

Great for coding and technical queries. Perfect for understanding code vulnerabilities and technical implementations.

How to Switch Models:

1. Click the model selector dropdown at the bottom of the chat interface
2. Browse available models with descriptions
3. Select your preferred model
4. Continue your conversation with the new model

Chat Interface Features

Message Input

Ask anything: Type your security questions in the message box at the bottom of the interface.

Supported Actions:

• Type text queries
• Press Enter to send
• Shift + Enter for new line
• Use formatting as needed

Input Enhancement Icons

Web Search (🌐)

Click to enable web search for the latest threat intelligence and security information from the internet.

Organization Database (📊)

Click to query your organization's security database for environment-specific insights.

Tools (🔧)

Click to access Developer & Security Tools including TraQez Scanners and other security utilities.

Response Actions

After receiving a response from DeepSec AI Agent, you can:

Copy Response

Click the Copy icon to copy the entire response to your clipboard for documentation or sharing.

Retry Query

Click the Retry icon to regenerate the response with the same question, useful if:

• The response wasn't satisfactory
• You want a different perspective
• There was an error in the initial response

Edit Message

Click the Edit icon to modify your original question and get a refined response.

Provide Feedback

Click the Feedback icon to rate the response quality and help improve the AI agent.

Create Note

Click the Note icon to save important responses as notes for future reference.

Regenerate

Click the Regenerate icon to get a new response using the same query.

Undo

Click the Undo icon to revert to the previous message state.

Developer & Security Tools

TraQez Scanners

Access advanced security scanning tools directly from the chat interface.

Description: Advanced security scanning using open source and deeptraq proprietary scanners.

Use Cases:

• Run on-demand vulnerability scans
• Integrate scanning into your workflow
• Access specialized security tools
• Automate security assessments

How to Access:

1. Click the Tools icon (🔧) in the message input area
2. Select Developer & Security Tools
3. Choose TraQez Scanners
4. Follow the prompts to configure and run scan

Chat Management

Starting a New Chat

Click New chat at the top of the sidebar to begin a fresh conversation with a clean context.

Searching Conversations

Use the Search Conversations box to find specific chats by:

• Keywords or topics
• CVE numbers
• Vulnerability names
• Date ranges

Organizing Chats

Naming Conversations: Chats are automatically titled based on the first message or question asked.

Chronological Order: Recent chats appear at the top of the list with timestamps.

Filtering: Search functionality helps find relevant conversations quickly.

Deleting Chats

Click the delete icon (🗑️) next to any chat to permanently remove it from your history.

Model Performance Comparison

Model	Speed	Accuracy	Best For
GPT-4o	Slow	Excellent	Complex analysis, multimodal
Deepseek Chat	Fast	Very Good	General security queries
Gemini 2.0-flash	Very Fast	Good	Quick answers, summaries
Llama 70b	Medium	Very Good	Detailed explanations
Deepseek 32b	Fast	Good	Technical/coding queries

Choosing the Right Model

Use GPT-4o when:

• You need the most comprehensive analysis
• Working with complex, multi-faceted security issues
• Requires deep reasoning and critical thinking

Use Deepseek Chat when:

• You need balanced speed and quality
• General security questions and analysis
• Default choice for most queries

Use Gemini 2.0-flash when:

• You need immediate responses
• Simple queries or quick summaries
• Time-sensitive situations

Use Llama 70b when:

• You want detailed, thorough explanations
• Open-source transparency is important
• Comprehensive security assessments

Use Deepseek 32b when:

• Analyzing code vulnerabilities
• Technical implementation questions
• Understanding exploit code

Integration with Scanning Tools

TraQez Scanners Integration

DeepSec AI Agent can trigger and analyze scans through the Developer & Security Tools integration.

Capabilities:

• Initiate vulnerability scans
• Schedule recurring assessments
• Analyze scan results
• Compare scan history
• Generate reports

How to Use:

1. Click the Tools icon in the chat interface
2. Select TraQez Scanners
3. Configure scan parameters
4. Review results within the conversation
5. Ask follow-up questions about findings

Best Practices

Effective Prompting

Do: ✓ Be specific with CVE numbers and asset details ✓ Provide context about your environment ✓ Ask one clear question at a time ✓ Use follow-up questions for clarity ✓ Include scan results when relevant

Don't: ✗ Share sensitive credentials or passwords ✗ Ask multiple unrelated questions at once ✗ Expect real-time system access ✗ Rely solely on AI for critical security decisions ✗ Share confidential business information

DeepSec AI Agent is continuously learning and improving based on user feedback and the latest security intelligence.