Skip to main content
Developer docs
core/safety

Cost Confidence Tiers

Tier A (verified cost), B (inferred), C (orphan). The tier on a SKU decides what the agent is allowed to do.

Cost-confidence tiers match the reversibility of an action to the confidence you have in its economics. The tier is stamped on the decision row at the moment of action, so an auditor sees the confidence the agent actually had — not what's known after costs were restated.

The tiers

  • Tier A — verified landed cost, current. Full action set including price changes.
  • Tier B — inferred margin within a confident band. Soft, reversible actions (copy, rank, draft).
  • Tier C — orphan / unknown cost. Observe and propose; archive needs operator confirm.

What each tier unlocks

tier → allowed actionspython
TIER_ACTIONS = {
    "A": {"PRICE_DROP", "DISCOUNT_TEST", "SCALE_WINNER", "OPTIMIZE_LOSER", "DRAFT", "VAULT"},
    "B": {"OPTIMIZE_LOSER", "DRAFT", "REVIVE_SEASONAL"},   # no price moves
    "C": {"FLAG_ORPHAN"},                                   # observe + propose only
}

Because tiers gate actions in the type system, margin policy is enforced at decision time rather than living in a doc nobody reads. See the Action enum for the full move set.

// developer docs

Magistry runs your store. We run Magistry.

The SDK is the same one our worker uses in production. Same auth, same rate limits, same audit plane — no shadow surface.

Stable for a year · Versioned changelogs · No silent breaking changes