Observed arrival · 2026-09-13
Scopebond puts a kill switch between AI agents and real-world damage
An open-source control layer that blocks out-of-policy AI-agent actions and records allowed or denied actions in tamper-evident receipts.
Field notes
The proposed architecture places a gateway between an agent and the systems it can touch, enforcing rules for spending, access, approvals, and resource changes. The homepage says it works with agents over HTTP or MCP and records both permitted and denied actions in tamper-evident receipts. Its live scenario uses an attempted $18,000 transfer against a $10,000 cap, giving visitors a concrete view of the blocking behavior without requiring signup.
Observed signals
Read the marks
Editorial observations of this landing page, not a rating.
One card from the complete issue