1. Introduction
Raypher ("we," "us," or "our") provides hardware-enforced security for the agentic era. We are
committed to a "Privacy by Architecture" philosophy, where security is enforced at the silicon
level without compromising your sensitive data.
2. Information We Collect
To provide AI Agent security, we collect data focused on execution integrity and threat patterns:
- TPM Attestation Data: Hardware hashes and PCR state to verify device
integrity.
- Agent Binary Metadata: Hashes and IDs of the agents you authorize Raypher
to protect.
- Kernel Event Logs: Anonymized syscall patterns (e.g., file access, network
requests) for security enforcement.
- DLP Indicators: Metadata regarding blocked PII/Secret exfiltration (we do
not store the secrets themselves).
3. Zero-Knowledge Guarantees
We DO NOT collect or have access to:
- AI Agent Prompts or completions (processed locally or via transparent proxy).
- Private Keys or Secrets (stored securely in your device's TPM/Hardware).
- The content of your redacted PII (DLP scanning happens at the runtime layer).
- Original Agent codebases (we only see the execution patterns).
4. Data Usage and Security
Data is used exclusively for:
- Enforcing your defined security policies.
- Improving our global Threat Intelligence Network (anonymized patterns).
- Providing visibility through the Cyberpunk Command Center dashboard.
5. Local-First Processing
The majority of Raypher's security logic, including kernel-level blocking and hardware
attestation, occurs locally on your machine. Cloud components are used only for multi-device
synchronization and pattern-wide threat updates.
6. Contact Us
For questions regarding our privacy architecture:
Email: privacy@raypherlabs.tech