00/Security
A compliance badge does not prove a system is secure. Reading the code does. Below is what ships today, what lands at launch in May 2026, and what's on the roadmap. Everything is tagged so you can tell the difference.
01/How we think
Principle / 01
Every security claim on this page is tagged today, launch, or roadmap. If we do not have something yet, we say so. That is the bar.
Principle / 02
Agent tokens are scoped and time-bound. Permissions are explicit. A leaked token blasts the smallest possible radius.
Principle / 03
The core library is MIT. Auditors, customers, and researchers can read the code. Nothing security-critical lives in a proprietary black box.
Principle / 04
Security incidents get a public incident report with timeline, impact, root cause, and fix. Published within seven days.
02/Infrastructure
The serverless choice is the security choice. Fewer things to forget to patch, no SSH doors to leave open, no long-lived credentials in motion. Every request terminates at the edge, runs, and ends.
Today
Today
Today
Today
At launch · May 2026
03/Data handling
Passwords are a burden we refuse. PII is a liability we minimize. Tokens that can be hashed, are hashed. If a breach happens, what leaks should be small by design, not small by luck.
Today
Today
Today
Today
At launch · May 2026
04/Compliance roadmap
Here is the real schedule. If one of these is a hard procurement requirement today, tell us up front and we will be honest about whether we can clear your gate in time.
Roadmap
Roadmap
At launch · May 2026
Roadmap
Roadmap
05/Coordinated disclosure
One business day response. No bounty yet. Researchers who help us ship a fix get a public write-up with credit and an early access seat.
Ground rules
If we can't say it today with a clean conscience, we won't put it on a security page.