AGENT ACTION FIREWALL · SCAN RESULT
Cloudflare
2 dangerous operation(s) in the documented tool surface; 2 can run without an accountable human receipt. 0 already gated.
WHAT CAN RUN WITHOUT A RECEIPT
✗ delete_dns_record — data_destruction
Add EMILIA Gate — @emilia-protocol/gate/adapters/supabase (or gateMcpTool) requiring a class_a receipt.
✗ purge_cache — data_destruction
Add EMILIA Gate — @emilia-protocol/gate/adapters/supabase (or gateMcpTool) requiring a class_a receipt.
Is this your project? Earn EG-1 Enforced.
Wrap the dangerous tools with @emilia-protocol/gate so they require a human/quorum receipt, re-run npx @emilia-protocol/fire-drill, and this result turns green. We update it on request.
Static assessment of the publicly-documented tool surface — not a live scan of any deployment and not a vulnerability report. It reflects whether the manifest requires a receipt for high-risk actions. Verify at runtime with EG-1 conformance.