P2C3: prove measured feedback over the World public journey - #6

Summary

• extend the released P2C2 official-record journey through exact observed Outcomes and domain-neutral measured feedback
• compare two reviewed exports of the exact LIVE Brief with two reviewed source-only controls under a frozen World-owned structural citation-coverage criterion
• append a useful evaluation and a non-effective, non-selectable promote proposal, then prove historical replay without reauthorization
• add candidate work-packet/audit evidence while leaving World 0.9.0, the Core 0.5 dependency window, Domain Pack JSON, and release records unchanged

Why

ACE Core 0.6 needs an external public-data consumer to falsify the neutral measured-impact contracts without moving World nouns or policy into Core. This packet is that first bounded source-checkout proof. It deliberately measures structural citation coverage only; it does not claim human benefit, causality, general Brief quality, network freshness, or effective promotion.

Dependency

Draft Core PR augmented-cognition-engine/core#88, currently at 9078018. With released ace-core==0.5.0, the two candidate tests skip explicitly and the public World boundary remains green.

Verification

Against Core PR #88 source plus the separately packaged reference action adapter:

• P2C3 focused: 2 passed
• complete World suite: 85 passed
• Federal Register adapter: 26 passed
• release contract: 7 passed
• focused Ruff and format checks: pass
• unchanged 0.9.0 wheel and sdist build: pass
• git diff --check: pass

Against the locked public Core 0.5 environment:

• P2C3 focused: 2 explicitly skipped
• complete World suite: 82 passed, 3 skipped

Remaining boundary

The proposal is not applied. The next packet is a separate human/Core-authorized accept-or-reject disposition. Broader World measures, opt-in network transport, and P2D multi-source correction remain independent work.