Security
Overview
AgentScope is designed with a focus on secure handling of AI observability data.
Data Protection
- Encryption in transit using HTTPS
- Secure storage using managed databases
- Logical data isolation per project
Access Control
- API key authentication
- Role-based access (planned)
- Restricted internal access
Infrastructure
- Hosted on secure cloud platforms
- Regular updates and patching
- Monitoring for suspicious activity
Sensitive Data
AgentScope may process:
- Prompts
- Model outputs
- Tool inputs/outputs
Users are responsible for avoiding:
- Secrets
- Credentials
- Highly sensitive personal data
Best Practices
We recommend:
- Masking sensitive data before sending
- Using test or synthetic data where possible
- Rotating API keys regularly
Incident Response
If a security issue is detected:
- We investigate immediately
- Affected users are notified if required