AI code governance under HELMtechnical

KAiM · TECHNICAL · DEMO DATA

AI code governance under HELM

An AI coding agent opens a risky pull request with a known-CVE dependency and missing human attestation. HELM blocks the merge and records the evidence a reviewer needs.

The six checks

Authority

Checked before the action executes.

Policy

Checked before the action executes.

Evidence

Checked before the action executes.

Harm

Checked before the action executes.

Regulation

Checked before the action executes.

Escalation

Checked before the action executes.

Outcome: HELM allows, blocks, or escalates the action and records a signed decision — no unauthorized customer-facing action executes, and any denial names a licensed human owner. See the gate →