diff --git a/tests/lifecycle/bats/electrumx.bats b/tests/lifecycle/bats/electrumx.bats
index 98ff38e6..05d7f2ff 100644
--- a/tests/lifecycle/bats/electrumx.bats
+++ b/tests/lifecycle/bats/electrumx.bats
@@ -27,6 +27,38 @@ teardown_file() {
   rpc_logout_local
 }
 
+# How far electrumx is behind its bitcoin daemon, from its own recent log
+# ("our height: N ... daemon: M"). Prints the gap; prints nothing when no
+# fresh (<30 min) sync line exists — callers must treat that as "unknown",
+# never as "synced".
+#
+# Why this exists: ElectrumX serves NO sessions until its initial sync has
+# caught up, and it flushes its DB cache at 1GB — i.e. rarely — so every
+# restart discards all unflushed progress back to the last flush. On
+# 2026-08-09 this node had spent 8d14h syncing largely because gate runs and
+# reboots kept taking its progress away. A gate that stop/start/restarts a
+# mid-sync electrumx therefore (a) honestly fails the serving probe and
+# (b) actively destroys hours of sync work. Skipping WITH THE GAP NAMED is
+# the truthful behaviour — this is a positively-detected syncing state, not
+# the container-absent skip trap fixed earlier in this file's history.
+electrumx_sync_gap() {
+  local line ours daemon
+  line=$(podman logs --tail 400 --since 30m electrumx 2>/dev/null \
+    | grep -E 'our height: [0-9,]+ daemon: [0-9,]+' | tail -1)
+  [[ -z "$line" ]] && return 0
+  ours=$(echo "$line" | grep -oE 'our height: [0-9,]+' | tr -dc '0-9')
+  daemon=$(echo "$line" | grep -oE 'daemon: [0-9,]+' | tr -dc '0-9')
+  [[ -n "$ours" && -n "$daemon" ]] && echo $((daemon - ours))
+}
+
+skip_if_initial_sync() {
+  local gap
+  gap=$(electrumx_sync_gap)
+  if [[ -n "$gap" ]] && (( gap > 10 )); then
+    skip "electrumx initial sync in progress ($gap blocks behind) — $1"
+  fi
+}
+
 # ────────────────────────────────────────────────────────────────────
 # Read-only tier
 # ────────────────────────────────────────────────────────────────────
@@ -52,6 +84,10 @@ teardown_file() {
     skip "electrumx not running (state=$state)"
   fi
 
+  # ElectrumX serves no sessions until initial sync completes, so probing a
+  # mid-sync instance can only fail — skip with the gap named instead.
+  skip_if_initial_sync "sessions are not served until it catches up"
+
   # Same probe required-stack.bats uses — divergence flags a real regression.
   # It is a real Electrum round-trip, not a bare connect(): podman's port
   # forwarder accepts the TCP handshake on the host-published port even when
@@ -96,6 +132,7 @@ PY
 
 @test "package.stop transitions electrumx to stopped" {
   [[ "${ARCHY_ALLOW_DESTRUCTIVE:-0}" == "1" ]] || skip "ARCHY_ALLOW_DESTRUCTIVE not set"
+  skip_if_initial_sync "restarting discards unflushed sync progress (1GB flush cache)"
 
   run rpc_result package.stop '{"id":"electrumx"}'
   [ "$status" -eq 0 ]
@@ -106,6 +143,7 @@ PY
 
 @test "package.start brings electrumx back to running" {
   [[ "${ARCHY_ALLOW_DESTRUCTIVE:-0}" == "1" ]] || skip "ARCHY_ALLOW_DESTRUCTIVE not set"
+  skip_if_initial_sync "restarting discards unflushed sync progress (1GB flush cache)"
 
   run rpc_result package.start '{"id":"electrumx"}'
   [ "$status" -eq 0 ]
@@ -116,6 +154,7 @@ PY
 
 @test "package.restart leaves electrumx in running state" {
   [[ "${ARCHY_ALLOW_DESTRUCTIVE:-0}" == "1" ]] || skip "ARCHY_ALLOW_DESTRUCTIVE not set"
+  skip_if_initial_sync "restarting discards unflushed sync progress (1GB flush cache)"
 
   run rpc_result package.restart '{"id":"electrumx"}'
   [ "$status" -eq 0 ]
@@ -126,6 +165,7 @@ PY
 
 @test "electrumx TCP port recovers after restart" {
   [[ "${ARCHY_ALLOW_DESTRUCTIVE:-0}" == "1" ]] || skip "ARCHY_ALLOW_DESTRUCTIVE not set"
+  skip_if_initial_sync "restarting discards unflushed sync progress (1GB flush cache)"
 
   # electrumx replays its index against bitcoind on cold start; allow 120s.
   local deadline=$(( $(date +%s) + 120 ))
diff --git a/tests/lifecycle/bats/required-stack.bats b/tests/lifecycle/bats/required-stack.bats
index a52662f4..4a199c70 100644
--- a/tests/lifecycle/bats/required-stack.bats
+++ b/tests/lifecycle/bats/required-stack.bats
@@ -50,6 +50,24 @@ skip_if_not_installed() {
   container_installed "$1" || skip "$1 not installed on this node"
 }
 
+# ElectrumX serves no sessions until its initial sync completes, so the
+# protocol probe below can only fail mid-sync. Skip with the gap NAMED (a
+# positively-detected state from a fresh log line, never inferred from
+# absence). Canonical rationale lives in electrumx.bats next to its twin.
+skip_if_electrumx_initial_sync() {
+  local line ours daemon
+  line=$(podman logs --tail 400 --since 30m electrumx 2>/dev/null \
+    | grep -E 'our height: [0-9,]+ daemon: [0-9,]+' | tail -1)
+  [[ -z "$line" ]] && return 0
+  ours=$(echo "$line" | grep -oE 'our height: [0-9,]+' | tr -dc '0-9')
+  daemon=$(echo "$line" | grep -oE 'daemon: [0-9,]+' | tr -dc '0-9')
+  [[ -n "$ours" && -n "$daemon" ]] || return 0
+  local gap=$((daemon - ours))
+  if (( gap > 10 )); then
+    skip "electrumx initial sync in progress ($gap blocks behind) — sessions are not served until it catches up"
+  fi
+}
+
 # The subset of required_containers actually installed on this node.
 installed_required_containers() {
   local c
@@ -135,6 +153,7 @@ bitcoin_json() {
 
 @test "electrumx answers the Electrum protocol (not just an open socket)" {
   skip_if_not_installed electrumx
+  skip_if_electrumx_initial_sync
   # A bare connect() to the HOST-published port proves nothing: podman's port
   # forwarder accepts the TCP handshake even when nothing inside the container
   # is listening. On 2026-08-09 this test was green while mempool-api was in a