{"id":"detector.score.v1","title":"Detector record, per rule","unit":"detections","chain_set":"eip155:8453 (Base)","window":"all time (fired, checked, confirmed, reversed); now (open); since the poisoning watch's first block (poisoning_adjacent)","definition":"For each detector rule: fired = detections the rule logged; open = of those, live now under the incident rule; checked = fires later re-derived from stored payments and receipts by the verification job; confirmed = checked and still held; reversed = checked and withdrawn, each with its reason stored; poisoning_adjacent = fires whose wallet received an address-poisoning transfer within seven days either side, counted only for fires inside the poisoning watch's window. The poisoning watch's own record (address_poisoning_loss) counts losses and how many two independent reads confirmed.","method":"Read from the detection log (table incidents), the verification records (table findings: verified_at, verification.ok, retired_reason), and the poisoning watch (tables poison_events, poison_losses). Published with every detections document, hourly.","excludes":["Fires we never re-derived: they count in fired but not in checked.","Any case we cannot bind to a stored record. Known cases replayed in a research note are listed by name with their source, not added to the counts."],"does_not_assert":["A false-positive rate: checked is a non-random subset of fired.","That an unchecked fire was right or wrong.","That a poisoning-adjacent fire was caused by the poisoning."],"source":"AFG-Pulse incident detectors","since":"2026-10-05","url":"https://agenticfinancegraph.com/def/detector.score.v1","ladder":"https://agenticfinancegraph.com/ai-agent-liveness-methodology-l0-l9-ladder","license":"https://creativecommons.org/licenses/by/4.0/","attribution":"Agentic Finance Graph — https://agenticfinancegraph.com"}