Deployed agents operate on handed‑over knowledge: a frozen policy only consults a briefing of consolidated items written before the stream begins. The world then changes while the store stays still—paths close, flags flip, price bands shift; each item was true at handover, and failure stems from staleness rather than ignorance.
We introduce ERRAND, treating revalidation as a priced errand: a recheck competes with the task it protects for the same scarce actions, funded only when the per‑action value of resolving a doubt exceeds a running wage. The errand index is single‑peaked, vanishing at both belief extremes, so certainty in either direction costs nothing; free en‑route receipts preserve on‑path knowledge, and repair writes a new version, never a deletion.
Under equal action budgets in two drifting tool‑use worlds, ERRAND clears every non‑oracle policy on the preregistered calibers, becoming the primary strategy in every setting and conditional at every binding budget, leading eager revalidation by 10.0 pp at the base cap.
Restraint wins: without a cap, ERRAND stops on its own, spending only 11.0% of steps, while uncapped eager revalidation consumes 70.7% of steps and still finishes 4.5 pp behind capped ERRAND. The margin appears where the briefing’s coverage is thinnest—the shadow price of long‑tail knowledge: a small, well‑priced budget beats a larger store that never rechecks.
Review