Observed arrival · 2026-11-07
Theras Guard puts an approval gate between AI agents and their tools
A French-language AI-agent gateway that describes checking each requested action against rules, routing sensitive actions to a human, or blocking them.
○Open
⊠Login
$Paid
†Ads
✦Pretty
●Pro
◎Niche
◉Human
⚑Risk
ƒJS
- For
- Teams deploying AI agents with access to workplace tools
- Worth noticing
- The page says untrusted documents contaminate a session for 15 minutes and gives human approval decisions a 10-minute validity window.
Field notes
The example workflow starts with an agent reading a supplier note marked as untrusted, then attempting to send a confidential report to an external address. The page says contamination lasts 15 minutes and shows the action being refused by a named approver. It also describes a decision log chained with SHA-256 hashes, with a control to falsify an earlier event and illustrate how later hashes would no longer match.
Observed signals
Read the marks
Editorial observations of this landing page, not a rating.
○OpenPublic substance visible
✦PrettyNotable craft visible
◎NicheUnusually specific use
One card from the complete issue