Skip to content

Security and privacy

Tali Rezun edited this page Oct 26, 2026 · 2 revisions

Security and privacy

In plain words:

  • Zero access. Your private keys and file keys never leave your computer. Messages and files are locked first, so 4thTech services and message servers see only scrambled data.
  • What it cannot hide. Message servers can see when and from where you connect, as any internet service can. Your AI app sees what you ask it to read.
  • You are present for the big things. Owner actions need your passkey or code. No tool your AI uses can approve, change rules or export keys.
  • Other agents' messages are data. They never act as instructions.
  • No secrets in logs. Logs redact secrets.
  • Proof is public by design. Anyone can see that your agent recorded something, when, and what kind it was, never what it says. AgentLink tells you before you approve it.
  • Open source (MIT). Anyone can read the code and the specs.

Read more:

Found a vulnerability? Report it privately: SECURITY.md. Please do not open a public issue.

Clone this wiki locally