Your codebase stays in your hands.
Cerin was architected from day one for developers and organizations who cannot compromise on IP security. We prioritize on-device execution, client-side encryption, and a strict zero-training policy.
Zero Training Policy
Your code, prompt inputs, git branches, and workspace metadata are never used to train AI models. When using Bring-Your-Own-Key (Anthropic, OpenAI, DeepSeek), your data is governed directly by your provider's zero-retention developer API agreements.
✓ Full support for 100% offline local LLMs via Ollama & LM Studio
Local AES-256-GCM Key Vault
All API keys and credentials entered into Cerin Desktop are encrypted on-disk using authenticated AES-256-GCM. Encryption keys are bound to your OS secure credential storage (DPAPI on Windows, Keychain on macOS, Secret Service on Linux).
✓ Zero API key transmission to Octanon Labs servers
Subprocess Isolation Sandbox
When Cerin Autopilot executes build commands or tests, it spawns an isolated pseudo-terminal (pty) child process confined strictly to the active workspace folder. High-risk filesystem mutations require visual diff approval.
✓ Destructive commands (e.g. rm -rf, sudo) require explicit approval
Binary Bytecode Integrity
Cerin's core agent engine does not ship as readable JavaScript. All core modules are compiled into V8 bytecode and validated at startup using cerin_shield.node, blocking memory injection and unauthorized code tampering.
✓ Cryptographic SHA-256 manifest verification on launch
Vulnerability Disclosure & Bug Bounty
Octanon Labs takes security vulnerabilities seriously. If you believe you have discovered a vulnerability in Cerin Desktop or Cerin Anywhere, please report it to our security team.
PGP key available upon request. Response within 24 hours.
We will not pursue legal action for good-faith security research adhering to our disclosure policy.