Skip to content

Campaign

Before opening a listing, decide what an acquisition would accomplish and how far this campaign may go. scavenger.plan_campaign@1 pins the desired result, existing inventory, hard criteria, weighted preferences, budget, deadline, region, evidence tier, travel/delivery semantics, disclosure policy, and operating mode.

The revision-3 finite feed observation also accepts a complete immutable campaign seed directly. Its admission retains and validates those facts before publishing the held Run; it neither invokes plan_campaign nor fabricates a completed planning Run. Missing budget, deadline, region or policy is not inferred. Explicit empty or not-assessed values are permitted only by the seed schema and establish no suitability or eligibility.

Subject profiles

Profile Evidence and judgment Revision-one consequence
Bazaar Item identity, condition, attributed seller claims, landed cost and bounded bargaining for Bazoš-style listings. Terms, one approved cash-on-delivery (COD) commitment, parcel inspection, dispute or refusal.
Technology Exact CPU, board, BIOS, cooling, memory, GPU, storage, case, PSU, firmware and connector compatibility. Bazaar consequence after fresh compatibility and evidence checks.
Property Transaction type, area, rooms, locality, route/amenity observations, contradictions and professional questions. Shortlist and diligence only: no contact, viewing, offer, reservation, deposit, signature or payment.

These are Scavenger-owned versioned rules, not new Compositions or source adapters. Technology can refine Bazaar; property rules cannot silently become parcel or hardware judgment. Profile selection supplies no source, credential, private anchor, money, or effect authority.

Watch observes and reports. Concierge prepares questions and offers for review. Bounded Autopilot acts only within its fully pinned envelope. Changing subject, ceilings, permitted disclosure, payment rail, evidence requirement, or deadline creates a successor campaign revision.

Preference within the boundary

A candidate may meet every hard gate and still be unattractive to the buyer. PreferenceRevision@1 can refine that soft ordering while retaining the reviewed field, admitted judgments, scope, provenance, and predecessor. It cannot repair a failed hard gate, widen budget or authority, change disclosure, or rewrite earlier rankings.

The declared learning mode determines whether a reviewed revision may become active for the next observation. Otherwise it stays a proposal awaiting explicit acceptance. Market makes that learning visible in the full candidate field.