diff --git a/CHANGELOG.md b/CHANGELOG.md
index 6c6f0382..24822d32 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -9,6 +9,22 @@ Pithead ships as **one product, one version** — the version lives in the top-l
 [`VERSION`](VERSION) file and every released image is tagged with it. Releases are cut
 per the process in [`docs/dev/releasing.md`](docs/dev/releasing.md).
 
+## [1.10.2] - 2026-07-21
+
+### Fixed
+
+- **The payout wallet no longer reports the whole stack unhealthy during its first scan (#718).**
+  With the genesis-scan default, `wallet-rpc`'s initial full-chain scan takes hours, and
+  monero-wallet-rpc doesn't answer its RPC while scanning — so the healthcheck flipped `unhealthy`
+  and could fire stack-health alerts for the entire scan, even though the wallet was working. The
+  healthcheck now tolerates an unreachable RPC while a first-scan marker is present (armed on wallet
+  creation, cleared the first time the RPC answers), so it stays healthy through the initial scan
+  and turns strict once caught up.
+- **Control-gate refusal messages no longer cite a closed, unrelated issue (#713).** The
+  "security-sensitive setting" and "destructive change" refusals pointed operators at #338 (a
+  closed Telegram-control issue) for "out-of-band approval"; they now just name the real path —
+  edit `config.json` on the host and run `./pithead apply`.
+
 ## [1.10.1] - 2026-07-21
 
 ### Fixed
diff --git a/VERSION b/VERSION
index 4dae2985..5ad2491c 100644
--- a/VERSION
+++ b/VERSION
@@ -1 +1 @@
-1.10.1
+1.10.2
diff --git a/build/dashboard/pyproject.toml b/build/dashboard/pyproject.toml
index 2d80a85e..f56d129d 100644
--- a/build/dashboard/pyproject.toml
+++ b/build/dashboard/pyproject.toml
@@ -7,7 +7,7 @@ name = "mining-dashboard"
 # Keep in lockstep with the top-level VERSION file — the single source of truth for the stack version
 # (#44). A shell test (tests/stack/run.sh) fails if these drift; the dashboard *displays* the version
 # from VERSION (baked in as PITHEAD_VERSION, #58), so this is packaging metadata only.
-version = "1.10.1"
+version = "1.10.2"
 description = "Monitoring dashboard and XvB switching engine for Pithead"
 readme = "README.md"
 requires-python = ">=3.11"
diff --git a/build/dashboard/uv.lock b/build/dashboard/uv.lock
index 56dc42c5..5c768638 100644
--- a/build/dashboard/uv.lock
+++ b/build/dashboard/uv.lock
@@ -782,7 +782,7 @@ wheels = [
 
 [[package]]
 name = "mining-dashboard"
-version = "1.10.1"
+version = "1.10.2"
 source = { editable = "." }
 dependencies = [
     { name = "aiofiles" },
diff --git a/build/monero/wallet-entrypoint.sh b/build/monero/wallet-entrypoint.sh
index cff963dd..2c9f00d5 100644
--- a/build/monero/wallet-entrypoint.sh
+++ b/build/monero/wallet-entrypoint.sh
@@ -11,6 +11,11 @@ set -eu
 
 WALLET_DIR="${WALLET_DIR:-/home/ubuntu/wallets}"
 WALLET_FILE="$WALLET_DIR/payout-wallet"
+# Marker (#718): touched when a wallet is first created, cleared by the healthcheck on the first
+# successful RPC. While it exists, an unreachable RPC means "still on the initial scan" — which for
+# the genesis default is HOURS, during which monero-wallet-rpc is single-threaded and won't answer.
+# It lives in the volume so it persists across container recreates until the scan actually finishes.
+SCAN_MARKER="$WALLET_DIR/.payout-scanning"
 GEN_JSON="${GEN_JSON:-/tmp/gen.json}" # tmpfs; holds the view key for the create-from-keys step only
 DAEMON_ADDRESS="${MONERO_NODE_HOST:-127.0.0.1}:${MONERO_RPC_PORT:-18081}"
 
@@ -76,6 +81,9 @@ if [ ! -f "$WALLET_FILE" ]; then
     height="$(resolve_scan_height)"
     [ -n "$height" ] || height=0
     echo "Creating view-only payout wallet at restore height $height (#381)..."
+    # Mark the initial scan (#718): the healthcheck tolerates an unreachable RPC while this exists
+    # (genesis scan is hours) and clears it on the first successful RPC.
+    touch "$SCAN_MARKER" 2>/dev/null || true
     # The view key lives ONLY in this tmpfs file, never on argv.
     write_gen_json "$height"
     # --generate-from-json creates + opens the wallet, then keeps serving the RPC.
diff --git a/build/monero/wallet-healthcheck.sh b/build/monero/wallet-healthcheck.sh
index 118a6b4f..d6381084 100644
--- a/build/monero/wallet-healthcheck.sh
+++ b/build/monero/wallet-healthcheck.sh
@@ -1,16 +1,37 @@
 #!/bin/sh
-# monero-wallet-rpc liveness check (#381).
+# monero-wallet-rpc health (#381, #718).
 #
-# Reads the RPC credentials from the container's environment instead of taking them as arguments,
-# so they are NOT baked into the compose `healthcheck.test` — where `docker inspect` could read
-# them (#90). A JSON-RPC `get_version` proves the server is up and answering; it succeeds during
-# the initial scan too, so the long start_period in compose tolerates first-run blockchain scanning
-# without flapping unhealthy.
-set -eu
+# RPC-up is the normal signal: a JSON-RPC get_version proves the server is answering. Credentials
+# come from the environment, not argv, so `docker inspect` can't read them (#90).
+#
+# But during the INITIAL payout scan (#718) monero-wallet-rpc is single-threaded and heads-down for
+# the whole scan — with the genesis default it is HOURS, and it refuses the RPC the entire time. A
+# plain RPC check flaps unhealthy after the 2m start_period and spams stack-health alerts for the
+# whole first scan. So: a marker file (`.payout-scanning`, written by the entrypoint on wallet
+# creation, living in the volume so it survives recreates) means "still on the first scan" — while
+# it exists, an unreachable RPC is treated as healthy. The first time the RPC answers, the scan has
+# caught up, so we clear the marker and are strict from then on (a later RPC failure is a real
+# fault, not scan tolerance).
+set -u
+
+WALLET_DIR="${WALLET_DIR:-/home/ubuntu/wallets}"
+SCAN_MARKER="$WALLET_DIR/.payout-scanning"
+
+rpc_up() {
+    curl -fsS --digest \
+        -u "${WALLET_RPC_USERNAME:-wallet}:${WALLET_RPC_PASSWORD:-}" \
+        -o /dev/null \
+        -H 'Content-Type: application/json' \
+        -d '{"jsonrpc":"2.0","id":"0","method":"get_version"}' \
+        http://localhost:18082/json_rpc
+}
+
+if rpc_up; then
+    # Caught up and answering — retire the initial-scan grace so future RPC failures read as real.
+    rm -f "$SCAN_MARKER" 2>/dev/null || true
+    exit 0
+fi
 
-curl -fsS --digest \
-    -u "${WALLET_RPC_USERNAME:-wallet}:${WALLET_RPC_PASSWORD:-}" \
-    -o /dev/null \
-    -H 'Content-Type: application/json' \
-    -d '{"jsonrpc":"2.0","id":"0","method":"get_version"}' \
-    http://localhost:18082/json_rpc
+# RPC silent. Healthy only if we're still on the first scan (marker present); else it's a fault.
+[ -f "$SCAN_MARKER" ] && exit 0
+exit 1
diff --git a/pithead b/pithead
index 7252e10b..096d1187 100755
--- a/pithead
+++ b/pithead
@@ -5075,11 +5075,11 @@ control_approval_gate() { # <staged-file>
     bad=$(printf '%s' "$porcelain" | awk -F'\t' 'NF' | cut -f2 | grep -cvxE "$editable_re" || true)
     if [ "${bad:-0}" -gt 0 ]; then
         hit=$(printf '%s' "$porcelain" | awk -F'\t' 'NF' | cut -f2 | grep -m1 -vxE "$editable_re" || true)
-        printf 'this change alters a security-sensitive setting (%s) that is not committable from the dashboard — out-of-band approval is tracked in #338. Apply it from the host with `%s apply`.' "${hit:-unparseable change row}" "$0"
+        printf 'this change alters a security-sensitive setting (%s) that is not committable from the dashboard. Edit config.json on the host and run `%s apply`.' "${hit:-unparseable change row}" "$0"
         return 1
     fi
     if printf '%s\n' "$porcelain" | grep -qE $'^DEST\t'; then
-        printf 'this change is destructive and cannot be committed from the dashboard — out-of-band approval is tracked in #338. Apply it from the host with `%s apply`.' "$0"
+        printf 'this change is destructive and cannot be committed from the dashboard. Edit config.json on the host and run `%s apply`.' "$0"
         return 1
     fi
     # Approved: echo the changed key NAMES so the commit's audit entry can record WHAT changed
diff --git a/tests/stack/run.sh b/tests/stack/run.sh
index 3fcd6cad..68071c0e 100755
--- a/tests/stack/run.sh
+++ b/tests/stack/run.sh
@@ -1105,6 +1105,34 @@ assert_eq "scan height: auto -> genesis 0 (full payout history)" "$(rsh auto)" "
 assert_eq "scan height: empty -> genesis 0" "$(rsh '')" "0"
 assert_eq "scan height: explicit block kept verbatim" "$(rsh 2500000)" "2500000"
 
+# Wallet healthcheck (#718): during the multi-hour genesis scan monero-wallet-rpc refuses the RPC,
+# so the check must tolerate an unreachable RPC WHILE the initial-scan marker is present, and turn
+# strict once the RPC first answers. Stub `curl` on PATH to be the RPC up/down control.
+HCBIN="$SANDBOX/hc-bin"
+HCDIR="$SANDBOX/hc-wallet"
+mkdir -p "$HCBIN" "$HCDIR"
+mk_curl() {
+    printf '#!/bin/sh\nexit %s\n' "$1" >"$HCBIN/curl"
+    chmod +x "$HCBIN/curl"
+}
+run_hc() { (
+    PATH="$HCBIN:$PATH" WALLET_DIR="$HCDIR" sh "$ROOT/build/monero/wallet-healthcheck.sh" >/dev/null 2>&1
+    echo $?
+); }
+# RPC down + marker present (mid initial scan) -> healthy (the whole point of #718).
+mk_curl 7
+: >"$HCDIR/.payout-scanning"
+assert_eq "healthcheck: RPC down but scanning -> healthy (#718)" "$(run_hc)" "0"
+# RPC up -> healthy AND the marker is retired (scan caught up; strict from now on).
+mk_curl 0
+assert_eq "healthcheck: RPC up -> healthy (#718)" "$(run_hc)" "0"
+if [ -f "$HCDIR/.payout-scanning" ]; then bad "healthcheck: RPC up clears the scan marker (#718)" "marker still present"; else ok "healthcheck: RPC up clears the scan marker (#718)"; fi
+# RPC down + NO marker (scan already finished once) -> unhealthy: a real fault, not scan tolerance.
+mk_curl 7
+assert_eq "healthcheck: RPC down after scan done -> unhealthy (#718)" "$(run_hc)" "1"
+# The entrypoint arms the marker on wallet creation so the grace applies from first boot.
+assert_contains "wallet-entrypoint touches the scan marker on create (#718)" "$(cat "$ROOT/build/monero/wallet-entrypoint.sh")" 'touch "$SCAN_MARKER"'
+
 echo "== unit: clock_sync_status (mining is time-sensitive) =="
 # doctor's NTP check classifies timedatectl's NTPSynchronized: yes→synced, no→unsynced, else unknown.
 CLKBIN="$SANDBOX/clk-bin"
@@ -4814,7 +4842,7 @@ assert_eq "destructive candidate previews destructive:true" "$(jq -r '.destructi
 printf '{"id":"%s","action":"commit","actor":"admin"}\n' "$UUID3" >"$REQS/$UUID3.json"
 run_pending >/dev/null
 assert_eq "destructive commit is refused" "$(jq -r '.status' "$RESULTS/$UUID3.json" 2>/dev/null)" "rejected"
-assert_contains "destructive refusal points at #338" "$(jq -r '.error' "$RESULTS/$UUID3.json" 2>/dev/null)" "#338"
+assert_contains "destructive refusal names the host apply path, not stale #338 (#713)" "$(jq -r '.error' "$RESULTS/$UUID3.json" 2>/dev/null)" "Edit config.json on the host"
 assert_eq "refused destructive commit did not touch config.json" "$(jq -r '.monero.clearnet_initial_sync // false' "$C/config.json")" "false"
 [ ! -f "$STAGED/$UUID3.json" ] && ok "refused destructive intent cleared from staged" || bad "refused destructive intent cleared from staged" "still staged"
 # A NON-destructive commit still proceeds (pool switch mini -> nano is INFO, not DEST).