It runs on your server
01The agent, its keys, and its AI usage all run on a server you set up. There's no Decenchro cloud holding your data, so there's nothing of yours for us to lose.
Security
Being able to check is the security.
Most AI platforms ask you to trust their logs. Decenchro is built so you don’t have to trust anyone’s: every risky action is checked before it runs, and written to a record that can’t be quietly edited, including by us.
Every action is owned
02Each risky action is tied to that one agent's identity before it runs. No shared account, no guessing who did what.
Checked before it runs
03The guard says allow, pause, or block before the action runs, not after. A pause stops the agent and waits for a person. Blocked actions never run, and the refusal is part of the record too.
A record that can't be rewritten
04Actions, decisions, and memories all go into one shared record. Changing history would mean rewriting a blockchain, so the audit rests on math, not on log files you have to trust.
Keys stay with you
05Model keys, bot tokens, and the agent's signing key live on your server and never pass through the console in the clear. The dashboard shows whether each is set, never the value.
Admins act, viewers watch
06The console is password-protected with secure sessions. Admins can launch, stop, and reconfigure agents; viewers see everything and can change nothing.
What leaves your server, and what doesn’t.
- AI requests
- your account, your key
- Safety checks
- checked before it runs
- Memory & decisions
- saved to the record
- Secrets & keys
- never leave your server
- Data sent to us
- none
Responsible disclosure
Found something? Tell us before you tell anyone else.
The code is open for anyone to read. If you find a security hole in the agent, the record, or the console, email us directly. We read every report.