Guardian
Concept
As the default member of Agent Team, Guardian cooperates with other agents through asynchronous requires_approval and get_approval processes to achieve efficient man-machine collaboration while ensuring system security.
It ensures user controllability of AI behavior by intelligent approval policy and LLM context analysis to identify potentially risky actions.
Key features
- π‘οΈ Intelligent risk identification and assessment
- π€ User-friendly approval interaction
- π― Flexible security policy configuration
- β‘ Efficient teamwork mechanism
Approve Strategies
Always: Enforces user approval modeNever: Automatic pass modeAuto-conservative: A conservative strategy that tends to ask for approvalAuto-permissive: A permissive policy that tends to pass automatically
Risk level
High risk (always): Actions that must be approvedMedium risk (maybe): an action that needs to be evaluatedLow risk (never): actions that can be performed directly
Typical High Risk Scenarios
Web Actions
- Form submission behavior
- Email sending operations
- Transaction/purchase behavior
Code Manipulations
- File system operations
- System information access
- Unauthorized operation
File Manipulations
- Suspicious file handling
- Configuration file modifications
Workflow
