The acquisition gate is an auditor with a positive-control, not a compliance blocker; it escalates to blocking only on logged evidence

Context

Every grounding gate today sits on the lossy publication side; acquisition — was the search well done? were the declared paid legs (x/exa) actually swept? — had no gate. A prior attempt at an enforcement gate on acquisition became "the monster": it optimized compliance, produced chokepoint→bypass→chokepoint (the #248 spiral), and fossilized rather than corrected. The operator's Voz (fragmentos-prosa-fiduciária): método livre, registro contratado — auditability that back-pressures acquisition because what is measured gets managed; and the loop closes through the operator, not over them.

Decision

Module 1 · Aquisição owns grounding.floor(manifest); Julgamento (Close) calls it, never re-implements it (or the format leak returns on the acquisition side).

The gate is an auditor: it records and surfaces the grounding manifest + yield (into the log, the briefing, the grill's agenda) and blocks nothing on thinness. The single enforcement exception is a positive-control on a "zero results" claim from a declared paid leg — the seca falsa, a false-negative from a broken instrument (the class an after-the-fact auditor cannot catch).

Escalation to a hard publish-block is a future decision made on logged evidence — pre-registered so the log already captures the deciding fields: the grill repeatedly catching thin acquisition, or a seca falsa passing the positive-control. Never on vibe. It is implemented today as the knob-driven harvest.close_floor (EDGE_GROUNDING_FLOOR: 0 off / 1 observe = auditor / 2 gate).

Consequences

• Método livre on depth; enforcement only on the canary — this is what avoids the monster

(compliance-optimization → bypass).

• The auditor→blocker move is an evidence-decided ratchet: logged, reversible, never silent.
• The loop closes through the operator — the grill sees thin acquisition and pushes; the gate does not

refuse.