The Capsule Security platform provides comprehensive AI agent discovery, monitoring, and security management across multiple platforms.
The platform integrates with leading AI agent platforms:
- Microsoft: CopilotStudio, PowerAutomate
- Salesforce: AgentForce
- ServiceNow: AgentSpace
- AWS: Bedrock
- Azure: Azure AI
- Google Cloud: Vertex AI
- Automation Tools: Zapier, UiPath, n8n
- Enterprise: Databricks, SAP, Box, Workato
Once integrated, the platform automatically discovers and catalogs:
- All deployed AI agents
- Agent configurations and permissions
- Tools and capabilities
- Data sources and connections
- Ownership information
- Environment details
For custom or unsupported platforms, agents can be manually registered through the API:
POST /api/v1/agents
{
"name": "Custom Support Agent",
"platform": "CUSTOM",
"environment": "production",
"owner": {
"email": "owner@example.com"
},
"accessibility": "TENANT",
"description": "Customer support automation agent",
"tools": ["email", "database", "api"]
}The agent list page provides powerful search and filtering:
GET /api/v1/agents?search=support&severity=HIGH&access=PUBLIC
// Filter parameters:
// - search: Text search in name and description
// - severity: LOW, MEDIUM, HIGH, CRITICAL
// - access: PUBLIC, TENANT, LIMITED
// - environment: production, staging, development
// - platform: COPILOTSTUDIO, POWERAUTOMATE, etc.
// - ownership: Owner email or ID
// - tools: Tool types used by agentThe AgentsList component displays agents in a responsive grid:
- Visual severity indicators
- Platform icons
- Owner information
- Last activity timestamps
- Quick access to details
- Bulk action capabilities
Each agent profile includes:
- Agent name and description
- Platform and environment
- Owner details with contact info
- Creation and modification dates
- Unique identifiers (internal and external)
Accessibility Level
- PUBLIC: Accessible to anyone
- TENANT: Restricted to tenant users
- LIMITED: Specific user access only
Risk Severity
- Highest issue severity indicator
- Number of open security issues
- Compliance status
Comprehensive list of tools the agent can access:
- Database connections
- API integrations
- File system access
- Email capabilities
- External service connections
Visual representation of:
- Data sources (nodes)
- Data connections (edges)
- Flow direction (ATOB, BTOA, BOTH)
- Security checkpoints
- Total session count
- Invocation frequency
- Last used timestamp
- Peak usage times
- User interaction patterns
Issues are categorized by type:
- Supply Chain: Third-party dependency risks
- Data: Data exposure or leakage risks
- Compliance: Regulatory compliance violations
- Integrity: Data integrity concerns
- Access: Unauthorized access risks
- Miscellaneous: Other security concerns
- CRITICAL: Immediate action required
- HIGH: Address within 24 hours
- MEDIUM: Address within 1 week
- LOW: Address in next maintenance cycle
- Detection: Automated scanning identifies issues
- Classification: Issues categorized and prioritized
- Assignment: Routed to appropriate team
- Resolution: Fix implemented and tested
- Verification: Issue confirmed resolved
GET /api/v1/agents/{agentId}/issues?status=OPEN
// Issue statuses:
// - OPEN: Awaiting resolution
// - RESOLVED: Issue fixed
// - IGNORED: Accepted riskThe platform provides continuous monitoring:
- Active session tracking
- Anomaly detection
- Performance metrics
- Error rate tracking
- Resource utilization
Visualize agent activity over time:
- Session frequency
- Duration patterns
- User interactions
- Peak usage periods
- Trend analysis
Ownership Assignment
- Every agent must have a designated owner
- Owners responsible for security compliance
- Regular ownership reviews
Access Control
- Follow principle of least privilege
- Regular access reviews
- Document access requirements
Tool Management
- Minimize tool permissions
- Regular capability audits
- Remove unused integrations
Regular Audits
- Weekly security issue reviews
- Monthly compliance checks
- Quarterly architecture reviews
Issue Resolution
- Prioritize CRITICAL and HIGH severity
- Document resolution steps
- Implement preventive measures
Monitoring
- Set up alerting for anomalies
- Track usage patterns
- Monitor for unauthorized access
Agents Not Appearing
- Verify integration is connected
- Check sync status
- Review platform permissions
- Wait for next sync cycle
Missing Agent Details
- Ensure platform API permissions
- Check network connectivity
- Review error logs
Incorrect Severity Ratings
- Refresh agent assessment
- Review issue detection rules
- Contact support for calibration