ARC-AI-002
High-impact tools have explicit active policy coverage
Pass
- Objective
- No tool that moves money, destroys data or changes privilege is reachable without a policy that deliberately governs it.
- Assurance
- automated — ARC decides this from its own records
- Frequency
- continuous
- Severity if it fails
- critical
- Enabled
- Yes
- Category
- authorization
Procedure
For every registered tool at risk tier high or critical, or declaring an irreversible effect, assert that at least one enabled policy selects it. A policy that matches everything by omitting selectors counts, and is reported as the reason, because a blanket policy is coverage but rarely the coverage anyone intended.
What a pass does not tell you
- Coverage is not correctness: a policy that matches the tool and allows everything satisfies this control.
- Risk tier and effect are declared by whoever registered the tool. A tool registered as low risk is assessed as low risk.
Latest result
- Status
- Pass
- Summary
- Every high-impact tool is governed by an enabled policy for all 2 active agents (2 pairs checked).
- Evaluated
- 2026-10-26 09:18:30Z
- Assessment
- arc_asm_59645e…
Covering 2026-09-26 09:18:30Z to 2026-10-26 09:18:30Z — 7 action(s) and 8 evidence record(s) fell in that window. Every status below describes that period and nothing outside it.
History
| Assessment | Status | Evaluated | Summary |
|---|---|---|---|
| arc_asm_59645e… | Pass | 2026-10-26 09:18:30Z | Every high-impact tool is governed by an enabled policy for all 2 active agents (2 pairs checked). |