PRODUCT / AI GOVERNANCE
Know exactly what your agents shipped.
Coding agents multiply commit volume. 'Who wrote this, and under what approval' stops being answerable by convention. ReARM makes it answerable by record.
01 / ATTRIBUTION
Per-commit attribution, cryptographically signed
Agent vs human, which agent, which model. Backed by signed commits, not commit-message convention.
PLACEHOLDER — SCREENSHOT 7: signed agent commit with trailers
02 / AGENT SESSIONS
Scoped, auditable work sessions
Each agent works in a session with orientation and final reports attached as evidence. What it was asked to do and what it says it did, on the record.
PLACEHOLDER — SCREENSHOT 7: agent session view
03 / POLICY GATES
The same policy engine that gates humans
Approval policies keyed on agent and session state. Agent-written code passes the same gates as human-written code, or it does not ship.
PLACEHOLDER — SCREENSHOT: policy gate on agent work
04 / ROLL-UP
Agent contribution, visible at every level
Per release, per component, per product: the same hierarchy as everything else. Agent work is not a separate system; it is a dimension of the one you already govern.
PLACEHOLDER — SCREENSHOT: agent contribution roll-up
WHERE THIS LIVES IN THE HIERARCHY
SBOM/xBOM Release Branch Component Feature Set Product