CAIN-INV-1204: Content inspection: find prompt injection, secrets and personal data in text an agent is about to read

VERIFIED source compiles

Domain	ADVERSARIAL_AND_INJECTION_DEFENSE
Tier	ENGINEERING_INNOVATION
Source	`sdk/python/src/cainstudio/inspection.py`
Mapped tests ran	108
Passed / failed / errored	108 / 0 / 0
Compiles	yes

Mapped tests

platform-gateway/test_acceptance_final.py

platform-gateway/tests/test_drift_security.py

platform-gateway/tests/test_phase4_sovereign_worker.py

sdk/python/tests/test_inspection.py

tests/govkernel/test_state.py

tests/test_cain42_e14_adversarial.py

Generated 2026-10-06T19:15:35Z from git unknown. This page is a republication of the run result; it is not an independent audit.

Same-project verification only. The tests are authored in this repository; no third party has reviewed or reproduced these results. VERIFIED means the mapped tests passed in this run, not that the mechanism is independently correct.