AI Security Assistant
The Redguard Portal features an integrated, real-time AI Security Assistant powered by Gemini. This persistent right-hand panel acts as an on-demand security consultant to analyze vulnerabilities, recommend fixes, and help you interact with your workspace using natural language.
Persistent AI Panel
The AI Assistant is accessible as a collapsible sidebar on almost all portal pages, or as a dedicated full-page workspace under /chat.
Key Capabilities
- Instant Explanations: Ask the assistant to explain complex vulnerabilities (e.g., "What is an insecure direct object reference?").
- Remediation Tailoring: Provide your back-end code (Ruby, Go, Python, etc.) and the assistant will write custom secure coding patches tailored to your specific findings.
- Asset Querying: Query your network inventory using natural language (e.g., "Show me all assets that have open critical findings").
- Interactive Shortcuts: Click on contextual buttons like "Find all open findings for asset 'PROD-DB-PRIMARY'" to launch immediate queries.
Human-in-the-Loop Validation
To maintain absolute control over your infrastructure and security catalog, the AI Assistant operates with a strict Human-in-the-Loop (HITL) validation model.
CAUTION
The AI Assistant can draft remediation plans or propose state changes, but it cannot modify database records directly without your explicit permission.
How HITL Works
- You ask the assistant to perform an action (e.g., "Set the SQL Injection finding to resolved").
- The assistant validates the command constraints and compiles a structured proposal card in the chat pane.
- A distinct Approve / Reject confirmation button appears below the proposal card.
- You review the proposal and click Approve to execute the database transaction, or click Reject to cancel.
Organization-Level Toggle
We recognize that some organizations operate under strict data residency and compliance guidelines that govern external LLM processing.
Disabling/Enabling the AI Assistant
- Administrative Toggle: Organization Administrators can toggle the AI Security Assistant on or off globally for their entire workspace.
- Settings Panel: Navigate to Settings > Organization Details to enable or disable LLM integrations.
- Privacy Assurance: When disabled, the right-hand panel will be completely deactivated for all members of the organization, and no data will be sent to Gemini APIs.