PromptShield
GenLayer network

Consensus security workbench / 01

Untrusted input. Canonical verdict.

Separate trusted intent from hostile content, then persist a validator-agreed prompt-injection and exfiltration assessment.

3 verdict bands10 attack categoriesGrounded evidence
01 / INPUT

Inspection envelope

0 / 8000
0 / 12000
0 / 2000
A wallet is required only for the contract write.
03 / LEDGER

Canonical history

Persisted classifications from this adapter.

0Total
0Safe
0Suspicious
0Dangerous

No persisted checks yet.