fix: address PR #905 review — migration error handling, backfill heuristic, test comment

1. Migration ALTER error no longer swallowed: check error from ALTER TABLE
   and return if it fails (unless column already exists). Migration is not
   marked complete on failure.

2. Backfill heuristic fixed: use observations table JOIN instead of
   packet_count > 0, since UpsertObserver sets packet_count = 1 on INSERT
   even for status-only observers.

3. Test clarifying comment: document that InsertTransmission uses
   data.Timestamp (not time.Now()) as source-of-truth for last_packet_at,
   so the hardcoded assertion is correct.

Dockerfile

@@ -14,6 +14,7 @@ WORKDIR /build/server
 COPY cmd/server/go.mod cmd/server/go.sum ./
 COPY internal/geofilter/ ../../internal/geofilter/
 COPY internal/sigvalidate/ ../../internal/sigvalidate/
+COPY internal/packetpath/ ../../internal/packetpath/
 RUN go mod download
 COPY cmd/server/ ./
 RUN CGO_ENABLED=0 GOOS=${TARGETOS} GOARCH=${TARGETARCH} \
@@ -24,6 +25,7 @@ WORKDIR /build/ingestor
 COPY cmd/ingestor/go.mod cmd/ingestor/go.sum ./
 COPY internal/geofilter/ ../../internal/geofilter/
 COPY internal/sigvalidate/ ../../internal/sigvalidate/
+COPY internal/packetpath/ ../../internal/packetpath/
 RUN go mod download
 COPY cmd/ingestor/ ./
 RUN CGO_ENABLED=0 GOOS=${TARGETOS} GOARCH=${TARGETARCH} \

cmd/ingestor/db.go

@@ -11,6 +11,7 @@ import (
 	"sync/atomic"
 	"time"
 
+	"github.com/meshcore-analyzer/packetpath"
 	_ "modernc.org/sqlite"
 )
 
@@ -112,7 +113,8 @@ func applySchema(db *sql.DB) error {
 			battery_mv INTEGER,
 			uptime_secs INTEGER,
 			noise_floor REAL,
-			inactive INTEGER DEFAULT 0
+			inactive INTEGER DEFAULT 0,
+			last_packet_at TEXT DEFAULT NULL
 		);
 
 		CREATE INDEX IF NOT EXISTS idx_nodes_last_seen ON nodes(last_seen);
@@ -189,7 +191,7 @@ func applySchema(db *sql.DB) error {
 	db.Exec(`DROP VIEW IF EXISTS packets_v`)
 	_, vErr := db.Exec(`
 		CREATE VIEW packets_v AS
-			SELECT o.id, t.raw_hex,
+			SELECT o.id, COALESCE(o.raw_hex, t.raw_hex) AS raw_hex,
 				   datetime(o.timestamp, 'unixepoch') AS timestamp,
 				   obs.id AS observer_id, obs.name AS observer_name,
 				   o.direction, o.snr, o.rssi, o.score, t.hash, t.route_type,
@@ -408,6 +410,37 @@ func applySchema(db *sql.DB) error {
 		log.Println("[migration] dropped_packets table created")
 	}
 
+	// Migration: add raw_hex column to observations (#881)
+	row = db.QueryRow("SELECT 1 FROM _migrations WHERE name = 'observations_raw_hex_v1'")
+	if row.Scan(&migDone) != nil {
+		log.Println("[migration] Adding raw_hex column to observations...")
+		db.Exec(`ALTER TABLE observations ADD COLUMN raw_hex TEXT`)
+		db.Exec(`INSERT INTO _migrations (name) VALUES ('observations_raw_hex_v1')`)
+		log.Println("[migration] observations.raw_hex column added")
+	}
+
+	// Migration: add last_packet_at column to observers (#last-packet-at)
+	row = db.QueryRow("SELECT 1 FROM _migrations WHERE name = 'observers_last_packet_at_v1'")
+	if row.Scan(&migDone) != nil {
+		log.Println("[migration] Adding last_packet_at column to observers...")
+		_, alterErr := db.Exec(`ALTER TABLE observers ADD COLUMN last_packet_at TEXT DEFAULT NULL`)
+		if alterErr != nil && !strings.Contains(alterErr.Error(), "duplicate column") {
+			return fmt.Errorf("observers last_packet_at ALTER: %w", alterErr)
+		}
+		// Backfill: set last_packet_at = last_seen only for observers that actually have
+		// observation rows (packet_count alone is unreliable — UpsertObserver sets it to 1
+		// on INSERT even for status-only observers).
+		res, err := db.Exec(`UPDATE observers SET last_packet_at = last_seen
+			WHERE last_packet_at IS NULL
+			AND rowid IN (SELECT DISTINCT observer_idx FROM observations WHERE observer_idx IS NOT NULL)`)
+		if err == nil {
+			n, _ := res.RowsAffected()
+			log.Printf("[migration] Backfilled last_packet_at for %d observers with packets", n)
+		}
+		db.Exec(`INSERT INTO _migrations (name) VALUES ('observers_last_packet_at_v1')`)
+		log.Println("[migration] observers.last_packet_at column added")
+	}
+
 	return nil
 }
 
@@ -433,12 +466,13 @@ func (s *Store) prepareStatements() error {
 	}
 
 	s.stmtInsertObservation, err = s.db.Prepare(`
-		INSERT INTO observations (transmission_id, observer_idx, direction, snr, rssi, score, path_json, timestamp)
-		VALUES (?, ?, ?, ?, ?, ?, ?, ?)
+		INSERT INTO observations (transmission_id, observer_idx, direction, snr, rssi, score, path_json, timestamp, raw_hex)
+		VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?)
 		ON CONFLICT(transmission_id, observer_idx, COALESCE(path_json, '')) DO UPDATE SET
-			snr   = COALESCE(excluded.snr,   snr),
-			rssi  = COALESCE(excluded.rssi,  rssi),
-			score = COALESCE(excluded.score, score)
+			snr     = COALESCE(excluded.snr,     snr),
+			rssi    = COALESCE(excluded.rssi,    rssi),
+			score   = COALESCE(excluded.score,   score),
+			raw_hex = COALESCE(excluded.raw_hex, raw_hex)
 	`)
 	if err != nil {
 		return err
@@ -490,7 +524,7 @@ func (s *Store) prepareStatements() error {
 		return err
 	}
 
-	s.stmtUpdateObserverLastSeen, err = s.db.Prepare("UPDATE observers SET last_seen = ? WHERE rowid = ?")
+	s.stmtUpdateObserverLastSeen, err = s.db.Prepare("UPDATE observers SET last_seen = ?, last_packet_at = ? WHERE rowid = ?")
 	if err != nil {
 		return err
 	}
@@ -569,9 +603,9 @@ func (s *Store) InsertTransmission(data *PacketData) (bool, error) {
 		err := s.stmtGetObserverRowid.QueryRow(data.ObserverID).Scan(&rowid)
 		if err == nil {
 			observerIdx = &rowid
-			// Update observer last_seen on every packet to prevent
+			// Update observer last_seen and last_packet_at on every packet to prevent
 			// low-traffic observers from appearing offline (#463)
-			_, _ = s.stmtUpdateObserverLastSeen.Exec(now, rowid)
+			_, _ = s.stmtUpdateObserverLastSeen.Exec(now, now, rowid)
 		}
 	}
 
@@ -584,7 +618,7 @@ func (s *Store) InsertTransmission(data *PacketData) (bool, error) {
 	_, err = s.stmtInsertObservation.Exec(
 		txID, observerIdx, data.Direction,
 		data.SNR, data.RSSI, data.Score,
-		data.PathJSON, epochTs,
+		data.PathJSON, epochTs, nilIfEmpty(data.RawHex),
 	)
 	if err != nil {
 		s.Stats.WriteErrors.Add(1)
@@ -931,11 +965,22 @@ type MQTTPacketMessage struct {
 }
 
 // BuildPacketData constructs a PacketData from a decoded packet and MQTT message.
+// path_json is derived directly from raw_hex header bytes (not decoded.Path.Hops)
+// to guarantee the stored path always matches the raw bytes. This matters for
+// TRACE packets where decoded.Path.Hops is overwritten with payload hops (#886).
 func BuildPacketData(msg *MQTTPacketMessage, decoded *DecodedPacket, observerID, region string) *PacketData {
 	now := time.Now().UTC().Format(time.RFC3339)
 	pathJSON := "[]"
-	if len(decoded.Path.Hops) > 0 {
-		b, _ := json.Marshal(decoded.Path.Hops)
+	// For TRACE packets, path_json must be the payload-decoded route hops
+	// (decoded.Path.Hops), NOT the raw_hex header bytes which are SNR values.
+	// For all other packet types, derive path from raw_hex (#886).
+	if !packetpath.PathBytesAreHops(byte(decoded.Header.PayloadType)) {
+		if len(decoded.Path.Hops) > 0 {
+			b, _ := json.Marshal(decoded.Path.Hops)
+			pathJSON = string(b)
+		}
+	} else if hops, err := packetpath.DecodePathFromRawHex(msg.Raw); err == nil && len(hops) > 0 {
+		b, _ := json.Marshal(hops)
 		pathJSON = string(b)
 	}
 

cmd/ingestor/db_test.go

@@ -2,6 +2,7 @@ package main
 
 import (
 	"database/sql"
+	"encoding/json"
 	"fmt"
 	"os"
 	"path/filepath"
@@ -10,6 +11,8 @@ import (
 	"sync/atomic"
 	"testing"
 	"time"
+
+	"github.com/meshcore-analyzer/packetpath"
 )
 
 func tempDBPath(t *testing.T) string {
@@ -566,6 +569,61 @@ func TestInsertTransmissionUpdatesObserverLastSeen(t *testing.T) {
 	}
 }
 
+func TestLastPacketAtUpdatedOnPacketOnly(t *testing.T) {
+	s, err := OpenStore(tempDBPath(t))
+	if err != nil {
+		t.Fatal(err)
+	}
+	defer s.Close()
+
+	// Insert observer via status path — last_packet_at should be NULL
+	if err := s.UpsertObserver("obs1", "Observer1", "SJC", nil); err != nil {
+		t.Fatal(err)
+	}
+
+	var lastPacketAt sql.NullString
+	s.db.QueryRow("SELECT last_packet_at FROM observers WHERE id = ?", "obs1").Scan(&lastPacketAt)
+	if lastPacketAt.Valid {
+		t.Fatalf("expected last_packet_at to be NULL after UpsertObserver, got %s", lastPacketAt.String)
+	}
+
+	// Insert a packet from this observer — last_packet_at should be set
+	data := &PacketData{
+		RawHex:      "0A00D69F",
+		Timestamp:   "2026-04-24T12:00:00Z",
+		ObserverID:  "obs1",
+		Hash:        "lastpackettest123456",
+		RouteType:   2,
+		PayloadType: 2,
+		PathJSON:    "[]",
+		DecodedJSON: `{"type":"TXT_MSG"}`,
+	}
+	if _, err := s.InsertTransmission(data); err != nil {
+		t.Fatal(err)
+	}
+
+	s.db.QueryRow("SELECT last_packet_at FROM observers WHERE id = ?", "obs1").Scan(&lastPacketAt)
+	if !lastPacketAt.Valid {
+		t.Fatal("expected last_packet_at to be non-NULL after InsertTransmission")
+	}
+	// InsertTransmission uses `now = data.Timestamp || time.Now()`, so last_packet_at
+	// should match the packet's Timestamp when provided (same source-of-truth as last_seen).
+	if lastPacketAt.String != "2026-04-24T12:00:00Z" {
+		t.Errorf("expected last_packet_at=2026-04-24T12:00:00Z, got %s", lastPacketAt.String)
+	}
+
+	// UpsertObserver again (status path) — last_packet_at should NOT change
+	if err := s.UpsertObserver("obs1", "Observer1", "SJC", nil); err != nil {
+		t.Fatal(err)
+	}
+
+	var lastPacketAtAfterStatus sql.NullString
+	s.db.QueryRow("SELECT last_packet_at FROM observers WHERE id = ?", "obs1").Scan(&lastPacketAtAfterStatus)
+	if !lastPacketAtAfterStatus.Valid || lastPacketAtAfterStatus.String != lastPacketAt.String {
+		t.Errorf("UpsertObserver should not change last_packet_at; expected %s, got %v", lastPacketAt.String, lastPacketAtAfterStatus)
+	}
+}
+
 func TestEndToEndIngest(t *testing.T) {
 	s, err := OpenStore(tempDBPath(t))
 	if err != nil {
@@ -1968,3 +2026,155 @@ func TestInsertObservationSNRFillIn(t *testing.T) {
 		t.Errorf("RSSI overwritten by null arrival: got %v, want %v", rssi3, rssi)
 	}
 }
+
+// TestPerObservationRawHex verifies that two MQTT packets for the same hash
+// from different observers store distinct raw_hex per observation (#881).
+func TestPerObservationRawHex(t *testing.T) {
+	store, err := OpenStore(tempDBPath(t))
+	if err != nil {
+		t.Fatal(err)
+	}
+	defer store.Close()
+
+	// Register two observers
+	store.UpsertObserver("obs-A", "Observer A", "", nil)
+	store.UpsertObserver("obs-B", "Observer B", "", nil)
+
+	hash := "abc123def456"
+	rawA := "c0ffee01"
+	rawB := "c0ffee0201aa"
+	dir := "RX"
+
+	// First observation from observer A
+	pdA := &PacketData{
+		RawHex:    rawA,
+		Hash:      hash,
+		Timestamp: "2026-04-21T10:00:00Z",
+		ObserverID: "obs-A",
+		Direction:  &dir,
+		PathJSON:   "[]",
+	}
+	isNew, err := store.InsertTransmission(pdA)
+	if err != nil {
+		t.Fatalf("insert A: %v", err)
+	}
+	if !isNew {
+		t.Fatal("expected new transmission")
+	}
+
+	// Second observation from observer B (same hash, different raw bytes)
+	pdB := &PacketData{
+		RawHex:    rawB,
+		Hash:      hash,
+		Timestamp: "2026-04-21T10:00:01Z",
+		ObserverID: "obs-B",
+		Direction:  &dir,
+		PathJSON:   `["aabb"]`,
+	}
+	isNew2, err := store.InsertTransmission(pdB)
+	if err != nil {
+		t.Fatalf("insert B: %v", err)
+	}
+	if isNew2 {
+		t.Fatal("expected duplicate transmission")
+	}
+
+	// Query observations and verify per-observation raw_hex
+	rows, err := store.db.Query(`
+		SELECT o.raw_hex, obs.id
+		FROM observations o
+		LEFT JOIN observers obs ON obs.rowid = o.observer_idx
+		ORDER BY o.id ASC
+	`)
+	if err != nil {
+		t.Fatalf("query: %v", err)
+	}
+	defer rows.Close()
+
+	type obsResult struct {
+		rawHex     string
+		observerID string
+	}
+	var results []obsResult
+	for rows.Next() {
+		var rh, oid sql.NullString
+		if err := rows.Scan(&rh, &oid); err != nil {
+			t.Fatal(err)
+		}
+		results = append(results, obsResult{
+			rawHex:     rh.String,
+			observerID: oid.String,
+		})
+	}
+
+	if len(results) != 2 {
+		t.Fatalf("expected 2 observations, got %d", len(results))
+	}
+	if results[0].rawHex != rawA {
+		t.Errorf("obs A raw_hex: got %q, want %q", results[0].rawHex, rawA)
+	}
+	if results[1].rawHex != rawB {
+		t.Errorf("obs B raw_hex: got %q, want %q", results[1].rawHex, rawB)
+	}
+	if results[0].rawHex == results[1].rawHex {
+		t.Error("both observations have same raw_hex — should differ")
+	}
+}
+
+// TestBuildPacketData_TraceUsesPayloadHops verifies that TRACE packets use
+// payload-decoded route hops in path_json (NOT the raw_hex header SNR bytes).
+// Issue #886 / #887.
+func TestBuildPacketData_TraceUsesPayloadHops(t *testing.T) {
+	// TRACE packet: header path has SNR bytes [30,2D,0D,23], but decoded.Path.Hops
+	// is overwritten to payload hops [67,33,D6,33,67].
+	rawHex := "2604302D0D2359FEE7B100000000006733D63367"
+	decoded, err := DecodePacket(rawHex, nil, false)
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	// decoded.Path.Hops should be the TRACE-replaced hops (payload hops)
+	if len(decoded.Path.Hops) != 5 {
+		t.Fatalf("expected 5 decoded hops, got %d", len(decoded.Path.Hops))
+	}
+
+	msg := &MQTTPacketMessage{Raw: rawHex}
+	pd := BuildPacketData(msg, decoded, "test-obs", "TST")
+
+	// For TRACE: path_json MUST be the payload-decoded route hops, NOT the SNR bytes
+	expectedPathJSON := `["67","33","D6","33","67"]`
+	if pd.PathJSON != expectedPathJSON {
+		t.Errorf("path_json = %s, want %s (TRACE must use payload hops)", pd.PathJSON, expectedPathJSON)
+	}
+
+	// Verify that DecodePathFromRawHex returns the SNR bytes (header path) which differ
+	headerHops, herr := packetpath.DecodePathFromRawHex(rawHex)
+	if herr != nil {
+		t.Fatal(herr)
+	}
+	headerJSON, _ := json.Marshal(headerHops)
+	if string(headerJSON) == expectedPathJSON {
+		t.Error("header path (SNR) should differ from payload hops for TRACE")
+	}
+}
+
+// TestBuildPacketData_NonTracePathJSON verifies non-TRACE packets also derive path from raw_hex.
+func TestBuildPacketData_NonTracePathJSON(t *testing.T) {
+	// A simple ADVERT packet (payload type 0) with 2 hops, hash_size 1
+	// Header 0x09 = FLOOD(1), ADVERT(2), version 0
+	// Path byte 0x02 = hash_size 1, hash_count 2
+	// Path bytes: AA BB
+	rawHex := "0902AABB" + "00000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000"
+	decoded, err := DecodePacket(rawHex, nil, false)
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	msg := &MQTTPacketMessage{Raw: rawHex}
+	pd := BuildPacketData(msg, decoded, "obs1", "TST")
+
+	expectedPathJSON := `["AA","BB"]`
+	if pd.PathJSON != expectedPathJSON {
+		t.Errorf("path_json = %s, want %s", pd.PathJSON, expectedPathJSON)
+	}
+}

cmd/ingestor/decoder.go

@@ -12,6 +12,7 @@ import (
 	"strings"
 	"unicode/utf8"
 
+	"github.com/meshcore-analyzer/packetpath"
 	"github.com/meshcore-analyzer/sigvalidate"
 )
 
@@ -192,8 +193,9 @@ func decodePath(pathByte byte, buf []byte, offset int) (Path, int) {
 	}, totalBytes
 }
 
+// isTransportRoute delegates to packetpath.IsTransportRoute.
 func isTransportRoute(routeType int) bool {
-	return routeType == RouteTransportFlood || routeType == RouteTransportDirect
+	return packetpath.IsTransportRoute(routeType)
 }
 
 func decodeEncryptedPayload(typeName string, buf []byte) Payload {

cmd/ingestor/decoder_test.go

@@ -11,6 +11,7 @@ import (
 	"strings"
 	"testing"
 
+	"github.com/meshcore-analyzer/packetpath"
 	"github.com/meshcore-analyzer/sigvalidate"
 )
 
@@ -1822,3 +1823,106 @@ func TestDecodeAdvertWithSignatureValidation(t *testing.T) {
 		t.Error("SignatureValid should be nil when validation disabled")
 	}
 }
+
+// === Tests for DecodePathFromRawHex (issue #886) ===
+
+func TestDecodePathFromRawHex_HashSize1(t *testing.T) {
+	// Header byte 0x26 = route_type DIRECT, payload TRACE
+	// Path byte 0x04 = hash_size 1 (bits 7-6 = 00 → 0+1=1), hash_count 4
+	// Path bytes: 30 2D 0D 23
+	raw := "2604302D0D2359FEE7B100000000006733D63367"
+	hops, err := packetpath.DecodePathFromRawHex(raw)
+	if err != nil {
+		t.Fatal(err)
+	}
+	expected := []string{"30", "2D", "0D", "23"}
+	if len(hops) != len(expected) {
+		t.Fatalf("got %d hops, want %d", len(hops), len(expected))
+	}
+	for i, h := range hops {
+		if h != expected[i] {
+			t.Errorf("hop[%d] = %s, want %s", i, h, expected[i])
+		}
+	}
+}
+
+func TestDecodePathFromRawHex_HashSize2(t *testing.T) {
+	// Path byte 0x42 = hash_size 2 (bits 7-6 = 01 → 1+1=2), hash_count 2
+	// Header 0x09 = FLOOD route (rt=1), payload ADVERT (pt=2)
+	// Path bytes: AABB CCDD (4 bytes = 2 hops * 2 bytes)
+	raw := "0942AABBCCDD" + "00000000000000"
+	hops, err := packetpath.DecodePathFromRawHex(raw)
+	if err != nil {
+		t.Fatal(err)
+	}
+	expected := []string{"AABB", "CCDD"}
+	if len(hops) != len(expected) {
+		t.Fatalf("got %d hops, want %d", len(hops), len(expected))
+	}
+	for i, h := range hops {
+		if h != expected[i] {
+			t.Errorf("hop[%d] = %s, want %s", i, h, expected[i])
+		}
+	}
+}
+
+func TestDecodePathFromRawHex_HashSize3(t *testing.T) {
+	// Path byte 0x81 = hash_size 3 (bits 7-6 = 10 → 2+1=3), hash_count 1
+	// Header 0x09 = FLOOD route (rt=1), payload ADVERT
+	raw := "0981AABBCC" + "0000000000"
+	hops, err := packetpath.DecodePathFromRawHex(raw)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if len(hops) != 1 || hops[0] != "AABBCC" {
+		t.Fatalf("got %v, want [AABBCC]", hops)
+	}
+}
+
+func TestDecodePathFromRawHex_HashSize4(t *testing.T) {
+	// Path byte 0xC1 = hash_size 4 (bits 7-6 = 11 → 3+1=4), hash_count 1
+	// Header 0x09 = FLOOD route (rt=1)
+	raw := "09C1AABBCCDD" + "0000000000"
+	hops, err := packetpath.DecodePathFromRawHex(raw)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if len(hops) != 1 || hops[0] != "AABBCCDD" {
+		t.Fatalf("got %v, want [AABBCCDD]", hops)
+	}
+}
+
+func TestDecodePathFromRawHex_DirectZeroHops(t *testing.T) {
+	// Path byte 0x00 = hash_size 1, hash_count 0
+	// Header 0x0A = DIRECT route (rt=2), payload ADVERT
+	raw := "0A00" + "0000000000"
+	hops, err := packetpath.DecodePathFromRawHex(raw)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if len(hops) != 0 {
+		t.Fatalf("got %d hops, want 0", len(hops))
+	}
+}
+
+func TestDecodePathFromRawHex_Transport(t *testing.T) {
+	// Route type 3 = TRANSPORT_DIRECT → 4 transport code bytes before path byte
+	// Header 0x27 = route_type 3, payload TRACE
+	// Transport codes: 1122 3344
+	// Path byte 0x02 = hash_size 1, hash_count 2
+	// Path bytes: AA BB
+	raw := "2711223344" + "02AABB" + "0000000000"
+	hops, err := packetpath.DecodePathFromRawHex(raw)
+	if err != nil {
+		t.Fatal(err)
+	}
+	expected := []string{"AA", "BB"}
+	if len(hops) != len(expected) {
+		t.Fatalf("got %d hops, want %d", len(hops), len(expected))
+	}
+	for i, h := range hops {
+		if h != expected[i] {
+			t.Errorf("hop[%d] = %s, want %s", i, h, expected[i])
+		}
+	}
+}

cmd/ingestor/go.mod

@@ -13,6 +13,10 @@ replace github.com/meshcore-analyzer/geofilter => ../../internal/geofilter
 
 replace github.com/meshcore-analyzer/sigvalidate => ../../internal/sigvalidate
 
+require github.com/meshcore-analyzer/packetpath v0.0.0
+
+replace github.com/meshcore-analyzer/packetpath => ../../internal/packetpath
+
 require (
 	github.com/dustin/go-humanize v1.0.1 // indirect
 	github.com/google/uuid v1.6.0 // indirect

cmd/server/bounded_load_test.go

@@ -229,7 +229,7 @@ func createTestDBAt(tb testing.TB, dbPath string, numTx int) {
 		id INTEGER PRIMARY KEY,
 		transmission_id INTEGER, observer_id TEXT, observer_name TEXT,
 		direction TEXT, snr REAL, rssi REAL, score INTEGER,
-		path_json TEXT, timestamp TEXT
+		path_json TEXT, timestamp TEXT, raw_hex TEXT
 	)`)
 	execOrFail(`CREATE TABLE IF NOT EXISTS observers (rowid INTEGER PRIMARY KEY, id TEXT, name TEXT)`)
 	execOrFail(`CREATE TABLE IF NOT EXISTS nodes (
@@ -280,7 +280,7 @@ func createTestDBWithObs(tb testing.TB, dbPath string, numTx int) {
 	)`)
 	execOrFail(`CREATE TABLE IF NOT EXISTS observations (
 		id INTEGER PRIMARY KEY, transmission_id INTEGER, observer_id TEXT, observer_name TEXT,
-		direction TEXT, snr REAL, rssi REAL, score INTEGER, path_json TEXT, timestamp TEXT
+		direction TEXT, snr REAL, rssi REAL, score INTEGER, path_json TEXT, timestamp TEXT, raw_hex TEXT
 	)`)
 	execOrFail(`CREATE TABLE IF NOT EXISTS observers (rowid INTEGER PRIMARY KEY, id TEXT, name TEXT)`)
 	execOrFail(`CREATE TABLE IF NOT EXISTS nodes (

cmd/server/coverage_test.go

@@ -47,7 +47,7 @@ func setupTestDBv2(t *testing.T) *DB {
 			id INTEGER PRIMARY KEY AUTOINCREMENT,
 			transmission_id INTEGER NOT NULL REFERENCES transmissions(id),
 			observer_id TEXT, observer_name TEXT, direction TEXT,
-			snr REAL, rssi REAL, score INTEGER, path_json TEXT, timestamp INTEGER NOT NULL
+			snr REAL, rssi REAL, score INTEGER, path_json TEXT, timestamp INTEGER NOT NULL, raw_hex TEXT
 		);
 	`
 	if _, err := conn.Exec(schema); err != nil {

cmd/server/db.go

@@ -20,6 +20,7 @@ type DB struct {
 	path             string // filesystem path to the database file
 	isV3             bool   // v3 schema: observer_idx in observations (vs observer_id in v2)
 	hasResolvedPath  bool   // observations table has resolved_path column
+	hasObsRawHex     bool   // observations table has raw_hex column (#881)
 
 	// Channel list cache (60s TTL) — avoids repeated GROUP BY scans (#762)
 	channelsCacheMu  sync.Mutex
@@ -76,6 +77,9 @@ func (db *DB) detectSchema() {
 			if colName == "resolved_path" {
 				db.hasResolvedPath = true
 			}
+			if colName == "raw_hex" {
+				db.hasObsRawHex = true
+			}
 		}
 	}
 }
@@ -166,6 +170,7 @@ type Observer struct {
 	BatteryMv     *int     `json:"battery_mv"`
 	UptimeSecs    *int64   `json:"uptime_secs"`
 	NoiseFloor    *float64 `json:"noise_floor"`
+	LastPacketAt  *string  `json:"last_packet_at"`
 }
 
 // Transmission represents a row from the transmissions table.
@@ -968,7 +973,7 @@ func (db *DB) getObservationsForTransmissions(txIDs []int) map[int][]map[string]
 
 // GetObservers returns all observers sorted by last_seen DESC.
 func (db *DB) GetObservers() ([]Observer, error) {
-	rows, err := db.conn.Query("SELECT id, name, iata, last_seen, first_seen, packet_count, model, firmware, client_version, radio, battery_mv, uptime_secs, noise_floor FROM observers ORDER BY last_seen DESC")
+	rows, err := db.conn.Query("SELECT id, name, iata, last_seen, first_seen, packet_count, model, firmware, client_version, radio, battery_mv, uptime_secs, noise_floor, last_packet_at FROM observers ORDER BY last_seen DESC")
 	if err != nil {
 		return nil, err
 	}
@@ -979,7 +984,7 @@ func (db *DB) GetObservers() ([]Observer, error) {
 		var o Observer
 		var batteryMv, uptimeSecs sql.NullInt64
 		var noiseFloor sql.NullFloat64
-		if err := rows.Scan(&o.ID, &o.Name, &o.IATA, &o.LastSeen, &o.FirstSeen, &o.PacketCount, &o.Model, &o.Firmware, &o.ClientVersion, &o.Radio, &batteryMv, &uptimeSecs, &noiseFloor); err != nil {
+		if err := rows.Scan(&o.ID, &o.Name, &o.IATA, &o.LastSeen, &o.FirstSeen, &o.PacketCount, &o.Model, &o.Firmware, &o.ClientVersion, &o.Radio, &batteryMv, &uptimeSecs, &noiseFloor, &o.LastPacketAt); err != nil {
 			continue
 		}
 		if batteryMv.Valid {
@@ -1002,8 +1007,8 @@ func (db *DB) GetObserverByID(id string) (*Observer, error) {
 	var o Observer
 	var batteryMv, uptimeSecs sql.NullInt64
 	var noiseFloor sql.NullFloat64
-	err := db.conn.QueryRow("SELECT id, name, iata, last_seen, first_seen, packet_count, model, firmware, client_version, radio, battery_mv, uptime_secs, noise_floor FROM observers WHERE id = ?", id).
-		Scan(&o.ID, &o.Name, &o.IATA, &o.LastSeen, &o.FirstSeen, &o.PacketCount, &o.Model, &o.Firmware, &o.ClientVersion, &o.Radio, &batteryMv, &uptimeSecs, &noiseFloor)
+	err := db.conn.QueryRow("SELECT id, name, iata, last_seen, first_seen, packet_count, model, firmware, client_version, radio, battery_mv, uptime_secs, noise_floor, last_packet_at FROM observers WHERE id = ?", id).
+		Scan(&o.ID, &o.Name, &o.IATA, &o.LastSeen, &o.FirstSeen, &o.PacketCount, &o.Model, &o.Firmware, &o.ClientVersion, &o.Radio, &batteryMv, &uptimeSecs, &noiseFloor, &o.LastPacketAt)
 	if err != nil {
 		return nil, err
 	}

cmd/server/db_test.go

@@ -48,7 +48,8 @@ func setupTestDB(t *testing.T) *DB {
 			radio TEXT,
 			battery_mv INTEGER,
 			uptime_secs INTEGER,
-			noise_floor REAL
+			noise_floor REAL,
+			last_packet_at TEXT DEFAULT NULL
 		);
 
 		CREATE TABLE transmissions (
@@ -74,7 +75,8 @@ func setupTestDB(t *testing.T) *DB {
 			score INTEGER,
 			path_json TEXT,
 			timestamp INTEGER NOT NULL,
-			resolved_path TEXT
+			resolved_path TEXT,
+			raw_hex TEXT
 		);
 
 		CREATE TABLE IF NOT EXISTS observer_metrics (
@@ -354,6 +356,10 @@ func TestGetObservers(t *testing.T) {
 	if observers[0].ID != "obs1" {
 		t.Errorf("expected obs1 first (most recent), got %s", observers[0].ID)
 	}
+	// last_packet_at should be nil since seedTestData doesn't set it
+	if observers[0].LastPacketAt != nil {
+		t.Errorf("expected nil LastPacketAt for obs1 from seed, got %v", *observers[0].LastPacketAt)
+	}
 }
 
 func TestGetObserverByID(t *testing.T) {
@@ -368,6 +374,48 @@ func TestGetObserverByID(t *testing.T) {
 	if obs.ID != "obs1" {
 		t.Errorf("expected obs1, got %s", obs.ID)
 	}
+	// Verify last_packet_at is nil by default
+	if obs.LastPacketAt != nil {
+		t.Errorf("expected nil LastPacketAt, got %v", *obs.LastPacketAt)
+	}
+}
+
+func TestGetObserverLastPacketAt(t *testing.T) {
+	db := setupTestDB(t)
+	defer db.Close()
+	seedTestData(t, db)
+
+	// Set last_packet_at for obs1
+	ts := "2026-04-24T12:00:00Z"
+	db.conn.Exec(`UPDATE observers SET last_packet_at = ? WHERE id = ?`, ts, "obs1")
+
+	// Verify via GetObservers
+	observers, err := db.GetObservers()
+	if err != nil {
+		t.Fatal(err)
+	}
+	var obs1 *Observer
+	for i := range observers {
+		if observers[i].ID == "obs1" {
+			obs1 = &observers[i]
+			break
+		}
+	}
+	if obs1 == nil {
+		t.Fatal("obs1 not found")
+	}
+	if obs1.LastPacketAt == nil || *obs1.LastPacketAt != ts {
+		t.Errorf("expected LastPacketAt=%s via GetObservers, got %v", ts, obs1.LastPacketAt)
+	}
+
+	// Verify via GetObserverByID
+	obs, err := db.GetObserverByID("obs1")
+	if err != nil {
+		t.Fatal(err)
+	}
+	if obs.LastPacketAt == nil || *obs.LastPacketAt != ts {
+		t.Errorf("expected LastPacketAt=%s via GetObserverByID, got %v", ts, obs.LastPacketAt)
+	}
 }
 
 func TestGetObserverByIDNotFound(t *testing.T) {
@@ -1108,7 +1156,8 @@ func setupTestDBV2(t *testing.T) *DB {
 			iata TEXT,
 			last_seen TEXT,
 			first_seen TEXT,
-			packet_count INTEGER DEFAULT 0
+			packet_count INTEGER DEFAULT 0,
+			last_packet_at TEXT DEFAULT NULL
 		);
 
 		CREATE TABLE transmissions (
@@ -1134,7 +1183,8 @@ func setupTestDBV2(t *testing.T) *DB {
 			rssi REAL,
 			score INTEGER,
 			path_json TEXT,
-			timestamp INTEGER NOT NULL
+			timestamp INTEGER NOT NULL,
+			raw_hex TEXT
 		);
 	`
 	if _, err := conn.Exec(schema); err != nil {
@@ -1975,3 +2025,59 @@ func TestParseWindowDuration(t *testing.T) {
 		}
 	}
 }
+
+// TestPerObservationRawHexEnrich verifies enrichObs returns per-observation raw_hex
+// when available, falling back to transmission raw_hex when NULL (#881).
+func TestPerObservationRawHexEnrich(t *testing.T) {
+	db := setupTestDB(t)
+	defer db.Close()
+
+	// Insert observers
+	db.conn.Exec(`INSERT INTO observers (id, name) VALUES ('obs-a', 'Observer A')`)
+	db.conn.Exec(`INSERT INTO observers (id, name) VALUES ('obs-b', 'Observer B')`)
+
+	var rowA, rowB int64
+	db.conn.QueryRow(`SELECT rowid FROM observers WHERE id='obs-a'`).Scan(&rowA)
+	db.conn.QueryRow(`SELECT rowid FROM observers WHERE id='obs-b'`).Scan(&rowB)
+
+	// Insert transmission with raw_hex
+	txHex := "deadbeef"
+	db.conn.Exec(`INSERT INTO transmissions (raw_hex, hash, first_seen) VALUES (?, 'hash1', '2026-04-21T10:00:00Z')`, txHex)
+
+	// Insert two observations: A has its own raw_hex, B has NULL (historical)
+	obsAHex := "c0ffee01"
+	db.conn.Exec(`INSERT INTO observations (transmission_id, observer_idx, snr, rssi, path_json, timestamp, raw_hex)
+		VALUES (1, ?, -5.0, -90.0, '[]', 1745236800, ?)`, rowA, obsAHex)
+	db.conn.Exec(`INSERT INTO observations (transmission_id, observer_idx, snr, rssi, path_json, timestamp)
+		VALUES (1, ?, -3.0, -85.0, '["aabb"]', 1745236801)`, rowB)
+
+	store := NewPacketStore(db, nil)
+	if err := store.Load(); err != nil {
+		t.Fatalf("store load: %v", err)
+	}
+
+	tx := store.byHash["hash1"]
+	if tx == nil {
+		t.Fatal("transmission not loaded")
+	}
+	if len(tx.Observations) < 2 {
+		t.Fatalf("expected 2 observations, got %d", len(tx.Observations))
+	}
+
+	// Check enriched observations
+	for _, obs := range tx.Observations {
+		m := store.enrichObs(obs)
+		rh, _ := m["raw_hex"].(string)
+		if obs.RawHex != "" {
+			// Observer A: should get per-observation raw_hex
+			if rh != obsAHex {
+				t.Errorf("obs with own raw_hex: got %q, want %q", rh, obsAHex)
+			}
+		} else {
+			// Observer B: should fall back to transmission raw_hex
+			if rh != txHex {
+				t.Errorf("obs without raw_hex: got %q, want %q (tx fallback)", rh, txHex)
+			}
+		}
+	}
+}

cmd/server/decoder.go

@@ -10,6 +10,7 @@ import (
 	"strings"
 	"time"
 
+	"github.com/meshcore-analyzer/packetpath"
 	"github.com/meshcore-analyzer/sigvalidate"
 )
 
@@ -164,8 +165,9 @@ func decodePath(pathByte byte, buf []byte, offset int) (Path, int) {
 	}, totalBytes
 }
 
+// isTransportRoute delegates to packetpath.IsTransportRoute.
 func isTransportRoute(routeType int) bool {
-	return routeType == RouteTransportFlood || routeType == RouteTransportDirect
+	return packetpath.IsTransportRoute(routeType)
 }
 
 func decodeEncryptedPayload(typeName string, buf []byte) Payload {
@@ -441,106 +443,6 @@ func DecodePacket(hexString string, validateSignatures bool) (*DecodedPacket, er
 	}, nil
 }
 
-// HexRange represents a labeled byte range for the hex breakdown visualization.
-type HexRange struct {
-	Start int    `json:"start"`
-	End   int    `json:"end"`
-	Label string `json:"label"`
-}
-
-// Breakdown holds colored byte ranges returned by the packet detail endpoint.
-type Breakdown struct {
-	Ranges []HexRange `json:"ranges"`
-}
-
-// BuildBreakdown computes labeled byte ranges for each section of a MeshCore packet.
-// The returned ranges are consumed by createColoredHexDump() and buildHexLegend()
-// in the frontend (public/app.js).
-func BuildBreakdown(hexString string) *Breakdown {
-	hexString = strings.ReplaceAll(hexString, " ", "")
-	hexString = strings.ReplaceAll(hexString, "\n", "")
-	hexString = strings.ReplaceAll(hexString, "\r", "")
-	buf, err := hex.DecodeString(hexString)
-	if err != nil || len(buf) < 2 {
-		return &Breakdown{Ranges: []HexRange{}}
-	}
-
-	var ranges []HexRange
-	offset := 0
-
-	// Byte 0: Header
-	ranges = append(ranges, HexRange{Start: 0, End: 0, Label: "Header"})
-	offset = 1
-
-	header := decodeHeader(buf[0])
-
-	// Bytes 1-4: Transport Codes (TRANSPORT_FLOOD / TRANSPORT_DIRECT only)
-	if isTransportRoute(header.RouteType) {
-		if len(buf) < offset+4 {
-			return &Breakdown{Ranges: ranges}
-		}
-		ranges = append(ranges, HexRange{Start: offset, End: offset + 3, Label: "Transport Codes"})
-		offset += 4
-	}
-
-	if offset >= len(buf) {
-		return &Breakdown{Ranges: ranges}
-	}
-
-	// Next byte: Path Length (bits 7-6 = hashSize-1, bits 5-0 = hashCount)
-	ranges = append(ranges, HexRange{Start: offset, End: offset, Label: "Path Length"})
-	pathByte := buf[offset]
-	offset++
-
-	hashSize := int(pathByte>>6) + 1
-	hashCount := int(pathByte & 0x3F)
-	pathBytes := hashSize * hashCount
-
-	// Path hops
-	if hashCount > 0 && offset+pathBytes <= len(buf) {
-		ranges = append(ranges, HexRange{Start: offset, End: offset + pathBytes - 1, Label: "Path"})
-	}
-	offset += pathBytes
-
-	if offset >= len(buf) {
-		return &Breakdown{Ranges: ranges}
-	}
-
-	payloadStart := offset
-
-	// Payload — break ADVERT into named sub-fields; everything else is one Payload range
-	if header.PayloadType == PayloadADVERT && len(buf)-payloadStart >= 100 {
-		ranges = append(ranges, HexRange{Start: payloadStart, End: payloadStart + 31, Label: "PubKey"})
-		ranges = append(ranges, HexRange{Start: payloadStart + 32, End: payloadStart + 35, Label: "Timestamp"})
-		ranges = append(ranges, HexRange{Start: payloadStart + 36, End: payloadStart + 99, Label: "Signature"})
-
-		appStart := payloadStart + 100
-		if appStart < len(buf) {
-			ranges = append(ranges, HexRange{Start: appStart, End: appStart, Label: "Flags"})
-			appFlags := buf[appStart]
-			fOff := appStart + 1
-			if appFlags&0x10 != 0 && fOff+8 <= len(buf) {
-				ranges = append(ranges, HexRange{Start: fOff, End: fOff + 3, Label: "Latitude"})
-				ranges = append(ranges, HexRange{Start: fOff + 4, End: fOff + 7, Label: "Longitude"})
-				fOff += 8
-			}
-			if appFlags&0x20 != 0 && fOff+2 <= len(buf) {
-				fOff += 2
-			}
-			if appFlags&0x40 != 0 && fOff+2 <= len(buf) {
-				fOff += 2
-			}
-			if appFlags&0x80 != 0 && fOff < len(buf) {
-				ranges = append(ranges, HexRange{Start: fOff, End: len(buf) - 1, Label: "Name"})
-			}
-		}
-	} else {
-		ranges = append(ranges, HexRange{Start: payloadStart, End: len(buf) - 1, Label: "Payload"})
-	}
-
-	return &Breakdown{Ranges: ranges}
-}
-
 // ComputeContentHash computes the SHA-256-based content hash (first 16 hex chars).
 // It hashes the payload-type nibble + payload (skipping path bytes) to produce a
 // route-independent identifier for the same logical packet. For TRACE packets,

cmd/server/decoder_test.go

@@ -97,146 +97,6 @@ func TestDecodePacket_FloodHasNoCodes(t *testing.T) {
 	}
 }
 
-func TestBuildBreakdown_InvalidHex(t *testing.T) {
-	b := BuildBreakdown("not-hex!")
-	if len(b.Ranges) != 0 {
-		t.Errorf("expected empty ranges for invalid hex, got %d", len(b.Ranges))
-	}
-}
-
-func TestBuildBreakdown_TooShort(t *testing.T) {
-	b := BuildBreakdown("11") // 1 byte — no path byte
-	if len(b.Ranges) != 0 {
-		t.Errorf("expected empty ranges for too-short packet, got %d", len(b.Ranges))
-	}
-}
-
-func TestBuildBreakdown_FloodNonAdvert(t *testing.T) {
-	// Header 0x15: route=1/FLOOD, payload=5/GRP_TXT
-	// PathByte 0x01: 1 hop, 1-byte hash
-	// PathHop: AA
-	// Payload: FF0011
-	b := BuildBreakdown("1501AAFFFF00")
-	labels := rangeLabels(b.Ranges)
-	expect := []string{"Header", "Path Length", "Path", "Payload"}
-	if !equalLabels(labels, expect) {
-		t.Errorf("expected labels %v, got %v", expect, labels)
-	}
-	// Verify byte positions
-	assertRange(t, b.Ranges, "Header", 0, 0)
-	assertRange(t, b.Ranges, "Path Length", 1, 1)
-	assertRange(t, b.Ranges, "Path", 2, 2)
-	assertRange(t, b.Ranges, "Payload", 3, 5)
-}
-
-func TestBuildBreakdown_TransportFlood(t *testing.T) {
-	// Header 0x14: route=0/TRANSPORT_FLOOD, payload=5/GRP_TXT
-	// TransportCodes: AABBCCDD (4 bytes)
-	// PathByte 0x01: 1 hop, 1-byte hash
-	// PathHop: EE
-	// Payload: FF00
-	b := BuildBreakdown("14AABBCCDD01EEFF00")
-	assertRange(t, b.Ranges, "Header", 0, 0)
-	assertRange(t, b.Ranges, "Transport Codes", 1, 4)
-	assertRange(t, b.Ranges, "Path Length", 5, 5)
-	assertRange(t, b.Ranges, "Path", 6, 6)
-	assertRange(t, b.Ranges, "Payload", 7, 8)
-}
-
-func TestBuildBreakdown_FloodNoHops(t *testing.T) {
-	// Header 0x15: FLOOD/GRP_TXT; PathByte 0x00: 0 hops; Payload: AABB
-	b := BuildBreakdown("150000AABB")
-	assertRange(t, b.Ranges, "Header", 0, 0)
-	assertRange(t, b.Ranges, "Path Length", 1, 1)
-	// No Path range since hashCount=0
-	for _, r := range b.Ranges {
-		if r.Label == "Path" {
-			t.Error("expected no Path range for zero-hop packet")
-		}
-	}
-	assertRange(t, b.Ranges, "Payload", 2, 4)
-}
-
-func TestBuildBreakdown_AdvertBasic(t *testing.T) {
-	// Header 0x11: FLOOD/ADVERT
-	// PathByte 0x01: 1 hop, 1-byte hash
-	// PathHop: AA
-	// Payload: 100 bytes (PubKey32 + Timestamp4 + Signature64) + Flags=0x02 (repeater, no extras)
-	pubkey := repeatHex("AB", 32)
-	ts := "00000000" // 4 bytes
-	sig := repeatHex("CD", 64)
-	flags := "02"
-	hex := "1101AA" + pubkey + ts + sig + flags
-	b := BuildBreakdown(hex)
-	assertRange(t, b.Ranges, "Header", 0, 0)
-	assertRange(t, b.Ranges, "Path Length", 1, 1)
-	assertRange(t, b.Ranges, "Path", 2, 2)
-	assertRange(t, b.Ranges, "PubKey", 3, 34)
-	assertRange(t, b.Ranges, "Timestamp", 35, 38)
-	assertRange(t, b.Ranges, "Signature", 39, 102)
-	assertRange(t, b.Ranges, "Flags", 103, 103)
-}
-
-func TestBuildBreakdown_AdvertWithLocation(t *testing.T) {
-	// flags=0x12: hasLocation bit set
-	pubkey := repeatHex("00", 32)
-	ts := "00000000"
-	sig := repeatHex("00", 64)
-	flags := "12" // 0x10 = hasLocation
-	latBytes := "00000000"
-	lonBytes := "00000000"
-	hex := "1101AA" + pubkey + ts + sig + flags + latBytes + lonBytes
-	b := BuildBreakdown(hex)
-	assertRange(t, b.Ranges, "Latitude", 104, 107)
-	assertRange(t, b.Ranges, "Longitude", 108, 111)
-}
-
-func TestBuildBreakdown_AdvertWithName(t *testing.T) {
-	// flags=0x82: hasName bit set
-	pubkey := repeatHex("00", 32)
-	ts := "00000000"
-	sig := repeatHex("00", 64)
-	flags := "82" // 0x80 = hasName
-	name := "4E6F6465" // "Node" in hex
-	hex := "1101AA" + pubkey + ts + sig + flags + name
-	b := BuildBreakdown(hex)
-	assertRange(t, b.Ranges, "Name", 104, 107)
-}
-
-// helpers
-
-func rangeLabels(ranges []HexRange) []string {
-	out := make([]string, len(ranges))
-	for i, r := range ranges {
-		out[i] = r.Label
-	}
-	return out
-}
-
-func equalLabels(a, b []string) bool {
-	if len(a) != len(b) {
-		return false
-	}
-	for i := range a {
-		if a[i] != b[i] {
-			return false
-		}
-	}
-	return true
-}
-
-func assertRange(t *testing.T, ranges []HexRange, label string, wantStart, wantEnd int) {
-	t.Helper()
-	for _, r := range ranges {
-		if r.Label == label {
-			if r.Start != wantStart || r.End != wantEnd {
-				t.Errorf("range %q: want [%d,%d], got [%d,%d]", label, wantStart, wantEnd, r.Start, r.End)
-			}
-			return
-		}
-	}
-	t.Errorf("range %q not found in %v", label, rangeLabels(ranges))
-}
 
 func TestZeroHopDirectHashSize(t *testing.T) {
 	// DIRECT (RouteType=2) + REQ (PayloadType=0) → header byte = 0x02

cmd/server/go.mod

@@ -14,6 +14,10 @@ replace github.com/meshcore-analyzer/geofilter => ../../internal/geofilter
 
 replace github.com/meshcore-analyzer/sigvalidate => ../../internal/sigvalidate
 
+require github.com/meshcore-analyzer/packetpath v0.0.0
+
+replace github.com/meshcore-analyzer/packetpath => ../../internal/packetpath
+
 require (
 	github.com/dustin/go-humanize v1.0.1 // indirect
 	github.com/google/uuid v1.6.0 // indirect

cmd/server/neighbor_persist_test.go

@@ -38,7 +38,7 @@ func createTestDBWithSchema(t *testing.T) (*DB, string) {
 		observer_id TEXT, observer_name TEXT, direction TEXT,
 		snr REAL, rssi REAL, score INTEGER,
 		path_json TEXT, timestamp TEXT,
-		resolved_path TEXT
+		resolved_path TEXT, raw_hex TEXT
 	)`)
 	conn.Exec(`CREATE TABLE nodes (
 		public_key TEXT PRIMARY KEY, name TEXT, role TEXT,
@@ -264,7 +264,7 @@ func TestEnsureResolvedPathColumn(t *testing.T) {
 	conn, _ := sql.Open("sqlite", "file:"+dbPath+"?_journal_mode=WAL")
 	conn.Exec(`CREATE TABLE observations (
 		id INTEGER PRIMARY KEY, transmission_id INTEGER,
-		observer_id TEXT, path_json TEXT, timestamp TEXT
+		observer_id TEXT, path_json TEXT, timestamp TEXT, raw_hex TEXT
 	)`)
 	conn.Close()
 

cmd/server/routes.go

@@ -16,6 +16,7 @@ import (
 	"time"
 
 	"github.com/gorilla/mux"
+	"github.com/meshcore-analyzer/packetpath"
 )
 
 // Server holds shared state for route handlers.
@@ -957,11 +958,9 @@ func (s *Server) handlePacketDetail(w http.ResponseWriter, r *http.Request) {
 		pathHops = []interface{}{}
 	}
 
-	rawHex, _ := packet["raw_hex"].(string)
 	writeJSON(w, PacketDetailResponse{
 		Packet:           packet,
 		Path:             pathHops,
-		Breakdown:        BuildBreakdown(rawHex),
 		ObservationCount: observationCount,
 		Observations:     mapSliceToObservations(observations),
 	})
@@ -1020,8 +1019,17 @@ func (s *Server) handlePostPacket(w http.ResponseWriter, r *http.Request) {
 
 	contentHash := ComputeContentHash(hexStr)
 	pathJSON := "[]"
-	if len(decoded.Path.Hops) > 0 {
-		if pj, e := json.Marshal(decoded.Path.Hops); e == nil {
+	// For TRACE packets, path_json must be the payload-decoded route hops
+	// (decoded.Path.Hops), NOT the raw_hex header bytes which are SNR values.
+	// For all other packet types, derive path from raw_hex (#886).
+	if !packetpath.PathBytesAreHops(byte(decoded.Header.PayloadType)) {
+		if len(decoded.Path.Hops) > 0 {
+			if pj, e := json.Marshal(decoded.Path.Hops); e == nil {
+				pathJSON = string(pj)
+			}
+		}
+	} else if hops, err := packetpath.DecodePathFromRawHex(hexStr); err == nil && len(hops) > 0 {
+		if pj, e := json.Marshal(hops); e == nil {
 			pathJSON = string(pj)
 		}
 	}

cmd/server/store.go

@@ -63,6 +63,7 @@ type StoreObs struct {
 	RSSI           *float64
 	Score          *int
 	PathJSON       string
+	RawHex         string
 	Timestamp      string
 }
 
@@ -458,6 +459,10 @@ func (s *PacketStore) Load() error {
 	if s.db.hasResolvedPath {
 		rpCol = ",\n\t\t\t\to.resolved_path"
 	}
+	obsRawHexCol := ""
+	if s.db.hasObsRawHex {
+		obsRawHexCol = ", o.raw_hex"
+	}
 
 	limitClause := ""
 	if maxPackets > 0 {
@@ -469,7 +474,7 @@ func (s *PacketStore) Load() error {
 		loadSQL = `SELECT t.id, t.raw_hex, t.hash, t.first_seen, t.route_type,
 				t.payload_type, t.payload_version, t.decoded_json,
 				o.id, obs.id, obs.name, o.direction,
-				o.snr, o.rssi, o.score, o.path_json, strftime('%Y-%m-%dT%H:%M:%fZ', o.timestamp, 'unixepoch')` + rpCol + `
+				o.snr, o.rssi, o.score, o.path_json, strftime('%Y-%m-%dT%H:%M:%fZ', o.timestamp, 'unixepoch')` + obsRawHexCol + rpCol + `
 			FROM transmissions t
 			LEFT JOIN observations o ON o.transmission_id = t.id
 			LEFT JOIN observers obs ON obs.rowid = o.observer_idx` + limitClause + `
@@ -478,7 +483,7 @@ func (s *PacketStore) Load() error {
 		loadSQL = `SELECT t.id, t.raw_hex, t.hash, t.first_seen, t.route_type,
 				t.payload_type, t.payload_version, t.decoded_json,
 				o.id, o.observer_id, o.observer_name, o.direction,
-				o.snr, o.rssi, o.score, o.path_json, o.timestamp` + rpCol + `
+				o.snr, o.rssi, o.score, o.path_json, o.timestamp` + obsRawHexCol + rpCol + `
 			FROM transmissions t
 			LEFT JOIN observations o ON o.transmission_id = t.id` + limitClause + `
 			ORDER BY t.first_seen ASC, o.timestamp DESC`
@@ -500,12 +505,16 @@ func (s *PacketStore) Load() error {
 		var observerID, observerName, direction, pathJSON, obsTimestamp sql.NullString
 		var snr, rssi sql.NullFloat64
 		var score sql.NullInt64
+		var obsRawHex sql.NullString
 		var resolvedPathStr sql.NullString
 
 		scanArgs := []interface{}{&txID, &rawHex, &hash, &firstSeen, &routeType, &payloadType,
 			&payloadVersion, &decodedJSON,
 			&obsID, &observerID, &observerName, &direction,
 			&snr, &rssi, &score, &pathJSON, &obsTimestamp}
+		if s.db.hasObsRawHex {
+			scanArgs = append(scanArgs, &obsRawHex)
+		}
 		if s.db.hasResolvedPath {
 			scanArgs = append(scanArgs, &resolvedPathStr)
 		}
@@ -565,6 +574,7 @@ func (s *PacketStore) Load() error {
 				RSSI:           nullFloatPtr(rssi),
 				Score:          nullIntPtr(score),
 				PathJSON:       obsPJ,
+				RawHex:         nullStrVal(obsRawHex),
 				Timestamp:      normalizeTimestamp(nullStrVal(obsTimestamp)),
 			}
 
@@ -1384,11 +1394,15 @@ func (s *PacketStore) IngestNewFromDB(sinceID, limit int) ([]map[string]interfac
 	// New ingests always resolve fresh using the current prefix map and neighbor graph.
 	// On restart, Load() handles reading persisted resolved_path values. (review item #7)
 	var querySQL string
+	obsRHCol := ""
+	if s.db.hasObsRawHex {
+		obsRHCol = ", o.raw_hex"
+	}
 	if s.db.isV3 {
 		querySQL = `SELECT t.id, t.raw_hex, t.hash, t.first_seen, t.route_type,
 				t.payload_type, t.payload_version, t.decoded_json,
 				o.id, obs.id, obs.name, o.direction,
-				o.snr, o.rssi, o.score, o.path_json, strftime('%Y-%m-%dT%H:%M:%fZ', o.timestamp, 'unixepoch')
+				o.snr, o.rssi, o.score, o.path_json, strftime('%Y-%m-%dT%H:%M:%fZ', o.timestamp, 'unixepoch')` + obsRHCol + `
 			FROM transmissions t
 			LEFT JOIN observations o ON o.transmission_id = t.id
 			LEFT JOIN observers obs ON obs.rowid = o.observer_idx
@@ -1398,7 +1412,7 @@ func (s *PacketStore) IngestNewFromDB(sinceID, limit int) ([]map[string]interfac
 		querySQL = `SELECT t.id, t.raw_hex, t.hash, t.first_seen, t.route_type,
 				t.payload_type, t.payload_version, t.decoded_json,
 				o.id, o.observer_id, o.observer_name, o.direction,
-				o.snr, o.rssi, o.score, o.path_json, o.timestamp
+				o.snr, o.rssi, o.score, o.path_json, o.timestamp` + obsRHCol + `
 			FROM transmissions t
 			LEFT JOIN observations o ON o.transmission_id = t.id
 			WHERE t.id > ?
@@ -1419,6 +1433,7 @@ func (s *PacketStore) IngestNewFromDB(sinceID, limit int) ([]map[string]interfac
 		routeType, payloadType                               *int
 		obsID                                                *int
 		observerID, observerName, direction, pathJSON, obsTS string
+		obsRawHex                                            string
 		snr, rssi                                            *float64
 		score                                                *int
 	}
@@ -1435,11 +1450,16 @@ func (s *PacketStore) IngestNewFromDB(sinceID, limit int) ([]map[string]interfac
 		var observerID, observerName, direction, pathJSON, obsTimestamp sql.NullString
 		var snrVal, rssiVal sql.NullFloat64
 		var scoreVal sql.NullInt64
+		var obsRawHex sql.NullString
 
-		if err := rows.Scan(&txID, &rawHex, &hash, &firstSeen, &routeType, &payloadType,
+		scanArgs2 := []interface{}{&txID, &rawHex, &hash, &firstSeen, &routeType, &payloadType,
 			&payloadVersion, &decodedJSON,
 			&obsIDVal, &observerID, &observerName, &direction,
-			&snrVal, &rssiVal, &scoreVal, &pathJSON, &obsTimestamp); err != nil {
+			&snrVal, &rssiVal, &scoreVal, &pathJSON, &obsTimestamp}
+		if s.db.hasObsRawHex {
+			scanArgs2 = append(scanArgs2, &obsRawHex)
+		}
+		if err := rows.Scan(scanArgs2...); err != nil {
 			continue
 		}
 
@@ -1464,6 +1484,7 @@ func (s *PacketStore) IngestNewFromDB(sinceID, limit int) ([]map[string]interfac
 			direction:    nullStrVal(direction),
 			pathJSON:     nullStrVal(pathJSON),
 			obsTS:        nullStrVal(obsTimestamp),
+			obsRawHex:    nullStrVal(obsRawHex),
 			snr:          nullFloatPtr(snrVal),
 			rssi:         nullFloatPtr(rssiVal),
 			score:        nullIntPtr(scoreVal),
@@ -1564,6 +1585,7 @@ func (s *PacketStore) IngestNewFromDB(sinceID, limit int) ([]map[string]interfac
 				RSSI:           r.rssi,
 				Score:          r.score,
 				PathJSON:       r.pathJSON,
+				RawHex:         r.obsRawHex,
 				Timestamp:      normalizeTimestamp(r.obsTS),
 			}
 
@@ -1806,9 +1828,13 @@ func (s *PacketStore) IngestNewObservations(sinceObsID, limit int) []map[string]
 	}
 
 	var querySQL string
+	obsRHCol2 := ""
+	if s.db.hasObsRawHex {
+		obsRHCol2 = ", o.raw_hex"
+	}
 	if s.db.isV3 {
 		querySQL = `SELECT o.id, o.transmission_id, obs.id, obs.name, o.direction,
-				o.snr, o.rssi, o.score, o.path_json, strftime('%Y-%m-%dT%H:%M:%fZ', o.timestamp, 'unixepoch')
+				o.snr, o.rssi, o.score, o.path_json, strftime('%Y-%m-%dT%H:%M:%fZ', o.timestamp, 'unixepoch')` + obsRHCol2 + `
 			FROM observations o
 			LEFT JOIN observers obs ON obs.rowid = o.observer_idx
 			WHERE o.id > ?
@@ -1816,7 +1842,7 @@ func (s *PacketStore) IngestNewObservations(sinceObsID, limit int) []map[string]
 			LIMIT ?`
 	} else {
 		querySQL = `SELECT o.id, o.transmission_id, o.observer_id, o.observer_name, o.direction,
-				o.snr, o.rssi, o.score, o.path_json, o.timestamp
+				o.snr, o.rssi, o.score, o.path_json, o.timestamp` + obsRHCol2 + `
 			FROM observations o
 			WHERE o.id > ?
 			ORDER BY o.id ASC
@@ -1839,6 +1865,7 @@ func (s *PacketStore) IngestNewObservations(sinceObsID, limit int) []map[string]
 		snr, rssi    *float64
 		score        *int
 		pathJSON     string
+		rawHex       string
 		timestamp    string
 	}
 
@@ -1848,9 +1875,14 @@ func (s *PacketStore) IngestNewObservations(sinceObsID, limit int) []map[string]
 		var observerID, observerName, direction, pathJSON, ts sql.NullString
 		var snr, rssi sql.NullFloat64
 		var score sql.NullInt64
+		var obsRawHex sql.NullString
 
-		if err := rows.Scan(&oid, &txID, &observerID, &observerName, &direction,
-			&snr, &rssi, &score, &pathJSON, &ts); err != nil {
+		scanArgs3 := []interface{}{&oid, &txID, &observerID, &observerName, &direction,
+			&snr, &rssi, &score, &pathJSON, &ts}
+		if s.db.hasObsRawHex {
+			scanArgs3 = append(scanArgs3, &obsRawHex)
+		}
+		if err := rows.Scan(scanArgs3...); err != nil {
 			continue
 		}
 
@@ -1864,6 +1896,7 @@ func (s *PacketStore) IngestNewObservations(sinceObsID, limit int) []map[string]
 			rssi:         nullFloatPtr(rssi),
 			score:        nullIntPtr(score),
 			pathJSON:     nullStrVal(pathJSON),
+			rawHex:       nullStrVal(obsRawHex),
 			timestamp:    nullStrVal(ts),
 		})
 	}
@@ -1919,6 +1952,7 @@ func (s *PacketStore) IngestNewObservations(sinceObsID, limit int) []map[string]
 			RSSI:           r.rssi,
 			Score:          r.score,
 			PathJSON:       r.pathJSON,
+			RawHex:         r.rawHex,
 			Timestamp:      normalizeTimestamp(r.timestamp),
 		}
 
@@ -2408,7 +2442,12 @@ func (s *PacketStore) enrichObs(obs *StoreObs) map[string]interface{} {
 
 	if tx != nil {
 		m["hash"] = strOrNil(tx.Hash)
-		m["raw_hex"] = strOrNil(tx.RawHex)
+		// Prefer per-observation raw_hex; fall back to transmission-level (#881)
+		if obs.RawHex != "" {
+			m["raw_hex"] = obs.RawHex
+		} else {
+			m["raw_hex"] = strOrNil(tx.RawHex)
+		}
 		m["payload_type"] = intPtrOrNil(tx.PayloadType)
 		m["route_type"] = intPtrOrNil(tx.RouteType)
 		m["decoded_json"] = strOrNil(tx.DecodedJSON)

cmd/server/types.go

@@ -315,7 +315,6 @@ type PacketTimestampsResponse struct {
 type PacketDetailResponse struct {
 	Packet           interface{}       `json:"packet"`
 	Path             []interface{}     `json:"path"`
-	Breakdown        *Breakdown        `json:"breakdown"`
 	ObservationCount int               `json:"observation_count"`
 	Observations     []ObservationResp `json:"observations,omitempty"`
 }

internal/packetpath/go.mod

@@ -0,0 +1,3 @@
+module github.com/meshcore-analyzer/packetpath
+
+go 1.22

internal/packetpath/path.go

@@ -0,0 +1,76 @@
+// Package packetpath provides shared helpers for extracting path hops from
+// raw MeshCore packet hex bytes.
+package packetpath
+
+import (
+	"encoding/hex"
+	"fmt"
+	"strings"
+)
+
+// DecodePathFromRawHex extracts the header path hops directly from raw hex bytes.
+// This is the authoritative path that matches what's in raw_hex, as opposed to
+// decoded.Path.Hops which may be overwritten for TRACE packets (issue #886).
+//
+// WARNING: This function returns the literal header path bytes regardless of
+// payload type. For TRACE packets these bytes are SNR values, NOT hop hashes.
+// Callers that may receive TRACE packets MUST check PathBytesAreHops(payloadType)
+// first, or use the safer DecodeHopsForPayload wrapper.
+func DecodePathFromRawHex(rawHex string) ([]string, error) {
+	buf, err := hex.DecodeString(rawHex)
+	if err != nil || len(buf) < 2 {
+		return nil, fmt.Errorf("invalid or too-short hex")
+	}
+
+	headerByte := buf[0]
+	offset := 1
+	if IsTransportRoute(int(headerByte & 0x03)) {
+		if len(buf) < offset+4 {
+			return nil, fmt.Errorf("too short for transport codes")
+		}
+		offset += 4
+	}
+	if offset >= len(buf) {
+		return nil, fmt.Errorf("too short for path byte")
+	}
+
+	pathByte := buf[offset]
+	offset++
+
+	hashSize := int(pathByte>>6) + 1
+	hashCount := int(pathByte & 0x3F)
+
+	hops := make([]string, 0, hashCount)
+	for i := 0; i < hashCount; i++ {
+		start := offset + i*hashSize
+		end := start + hashSize
+		if end > len(buf) {
+			break
+		}
+		hops = append(hops, strings.ToUpper(hex.EncodeToString(buf[start:end])))
+	}
+	return hops, nil
+}
+
+// DecodeHopsForPayload returns the header path hops only when the payload type's
+// header bytes are actually route hops (i.e. PathBytesAreHops(payloadType) is true).
+// For TRACE packets it returns (nil, ErrPayloadHasNoHeaderHops) so the caller is
+// forced to source hops from the decoded payload instead.
+//
+// Prefer this over DecodePathFromRawHex when the payload type is known.
+func DecodeHopsForPayload(rawHex string, payloadType byte) ([]string, error) {
+	if !PathBytesAreHops(payloadType) {
+		return nil, ErrPayloadHasNoHeaderHops
+	}
+	return DecodePathFromRawHex(rawHex)
+}
+
+// ErrPayloadHasNoHeaderHops is returned by DecodeHopsForPayload when the
+// payload type repurposes the raw_hex header path bytes (e.g. TRACE → SNR values).
+var ErrPayloadHasNoHeaderHops = errPayloadHasNoHeaderHops{}
+
+type errPayloadHasNoHeaderHops struct{}
+
+func (errPayloadHasNoHeaderHops) Error() string {
+	return "payload type repurposes header path bytes; source hops from decoded payload"
+}

internal/packetpath/path_test.go

@@ -0,0 +1,150 @@
+package packetpath
+
+import (
+	"encoding/hex"
+	"encoding/json"
+	"strings"
+	"testing"
+)
+
+func TestDecodePathFromRawHex_Basic(t *testing.T) {
+	// Build a simple FLOOD packet (route_type=1) with 2 hops of hashSize=1
+	// header: route_type=1, payload_type=2 (TXT_MSG), version=0 → 0b00_0010_01 = 0x09
+	// path byte: hashSize=1 (bits 7-6 = 0), hashCount=2 (bits 5-0 = 2) → 0x02
+	// hops: AB, CD
+	// payload: some bytes
+	raw := "0902ABCD" + "DEADBEEF"
+	hops, err := DecodePathFromRawHex(raw)
+	if err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+	if len(hops) != 2 || hops[0] != "AB" || hops[1] != "CD" {
+		t.Fatalf("expected [AB, CD], got %v", hops)
+	}
+}
+
+func TestDecodePathFromRawHex_ZeroHops(t *testing.T) {
+	// DIRECT route (type=2), no hops → 0b00_0010_10 = 0x0A
+	// path byte: 0x00 (0 hops)
+	raw := "0A00" + "DEADBEEF"
+	hops, err := DecodePathFromRawHex(raw)
+	if err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+	if len(hops) != 0 {
+		t.Fatalf("expected 0 hops, got %v", hops)
+	}
+}
+
+func TestDecodePathFromRawHex_TransportRoute(t *testing.T) {
+	// TRANSPORT_FLOOD (route_type=0), payload_type=5 (GRP_TXT), version=0
+	// header: 0b00_0101_00 = 0x14
+	// transport codes: 4 bytes
+	// path byte: hashSize=1, hashCount=1 → 0x01
+	// hop: FF
+	raw := "14" + "00112233" + "01" + "FF" + "DEAD"
+	hops, err := DecodePathFromRawHex(raw)
+	if err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+	if len(hops) != 1 || hops[0] != "FF" {
+		t.Fatalf("expected [FF], got %v", hops)
+	}
+}
+
+// buildTracePacket creates a TRACE packet hex string where header path bytes are
+// SNR values, and payload contains the actual route hops.
+func buildTracePacket() (rawHex string, headerPathHops []string, payloadHops []string) {
+	// DIRECT route (type=2), TRACE payload (type=9), version=0
+	// header byte: 0b00_1001_10 = 0x26
+	headerByte := byte(0x26)
+
+	// Header path: 2 SNR bytes (hashSize=1, hashCount=2) → path byte = 0x02
+	// SNR values: 0x1A (26 dB), 0x0F (15 dB)
+	pathByte := byte(0x02)
+	snrBytes := []byte{0x1A, 0x0F}
+
+	// TRACE payload: tag(4) + authCode(4) + flags(1) + path hops
+	tag := []byte{0x01, 0x00, 0x00, 0x00}
+	authCode := []byte{0x02, 0x00, 0x00, 0x00}
+	// flags: path_sz=0 (1 byte hops), other bits=0 → 0x00
+	flags := byte(0x00)
+	// Payload hops: AA, BB, CC (the actual route)
+	payloadPathBytes := []byte{0xAA, 0xBB, 0xCC}
+
+	var buf []byte
+	buf = append(buf, headerByte, pathByte)
+	buf = append(buf, snrBytes...)
+	buf = append(buf, tag...)
+	buf = append(buf, authCode...)
+	buf = append(buf, flags)
+	buf = append(buf, payloadPathBytes...)
+
+	rawHex = strings.ToUpper(hex.EncodeToString(buf))
+	headerPathHops = []string{"1A", "0F"} // SNR values — NOT route hops
+	payloadHops = []string{"AA", "BB", "CC"} // actual route hops from payload
+	return
+}
+
+func TestDecodePathFromRawHex_TraceReturnsSNR(t *testing.T) {
+	rawHex, expectedSNR, _ := buildTracePacket()
+	hops, err := DecodePathFromRawHex(rawHex)
+	if err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+	// DecodePathFromRawHex always returns header path bytes — for TRACE these are SNR values
+	if len(hops) != len(expectedSNR) {
+		t.Fatalf("expected %d hops (SNR), got %d: %v", len(expectedSNR), len(hops), hops)
+	}
+	for i, h := range hops {
+		if h != expectedSNR[i] {
+			t.Errorf("hop[%d]: expected %s, got %s", i, expectedSNR[i], h)
+		}
+	}
+}
+
+func TestTracePathJSON_UsesPayloadHops(t *testing.T) {
+	// This test validates the TRACE vs non-TRACE logic that callers should implement:
+	// For TRACE: path_json = decoded.Path.Hops (payload-decoded route hops)
+	// For non-TRACE: path_json = DecodePathFromRawHex(raw_hex)
+	rawHex, snrHops, payloadHops := buildTracePacket()
+
+	// DecodePathFromRawHex returns SNR bytes for TRACE
+	headerHops, _ := DecodePathFromRawHex(rawHex)
+	headerJSON, _ := json.Marshal(headerHops)
+
+	// payload hops (what decoded.Path.Hops would return after TRACE decoding)
+	payloadJSON, _ := json.Marshal(payloadHops)
+
+	// They must differ — SNR != route hops
+	if string(headerJSON) == string(payloadJSON) {
+		t.Fatalf("SNR hops and payload hops should differ for TRACE; both are %s", headerJSON)
+	}
+
+	// For TRACE, path_json should be payloadHops, not headerHops
+	_ = snrHops // snrHops == headerHops — used for documentation
+	t.Logf("TRACE: header path (SNR) = %s, payload path (route) = %s", headerJSON, payloadJSON)
+}
+
+func TestDecodeHopsForPayload_NonTrace(t *testing.T) {
+	// header 0x01, path_len 0x02, hops 0xAA 0xBB, then payload bytes
+	raw := "0102AABB00"
+	hops, err := DecodeHopsForPayload(raw, 0x05) // GRP_TXT — header path bytes ARE hops
+	if err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+	if len(hops) != 2 || hops[0] != "AA" || hops[1] != "BB" {
+		t.Errorf("expected [AA BB], got %v", hops)
+	}
+}
+
+func TestDecodeHopsForPayload_TraceReturnsError(t *testing.T) {
+	raw := "010205F00100"
+	hops, err := DecodeHopsForPayload(raw, PayloadTRACE)
+	if err != ErrPayloadHasNoHeaderHops {
+		t.Errorf("expected ErrPayloadHasNoHeaderHops, got %v", err)
+	}
+	if hops != nil {
+		t.Errorf("expected nil hops for TRACE, got %v", hops)
+	}
+}

internal/packetpath/route.go

@@ -0,0 +1,24 @@
+package packetpath
+
+// Route type constants (header bits 1-0).
+const (
+	RouteTransportFlood  = 0
+	RouteFlood           = 1
+	RouteDirect          = 2
+	RouteTransportDirect = 3
+)
+
+// PayloadTRACE is the payload type constant for TRACE packets.
+const PayloadTRACE = 0x09
+
+// IsTransportRoute returns true for TRANSPORT_FLOOD (0) and TRANSPORT_DIRECT (3).
+func IsTransportRoute(routeType int) bool {
+	return routeType == RouteTransportFlood || routeType == RouteTransportDirect
+}
+
+// PathBytesAreHops returns true when the raw_hex header path bytes represent
+// route hop hashes (the normal case). Returns false for packet types where
+// header path bytes are repurposed (e.g. TRACE uses them for SNR values).
+func PathBytesAreHops(payloadType byte) bool {
+	return payloadType != PayloadTRACE
+}

internal/packetpath/route_test.go

@@ -0,0 +1,31 @@
+package packetpath
+
+import "testing"
+
+func TestIsTransportRoute(t *testing.T) {
+	if !IsTransportRoute(RouteTransportFlood) {
+		t.Error("RouteTransportFlood should be transport")
+	}
+	if !IsTransportRoute(RouteTransportDirect) {
+		t.Error("RouteTransportDirect should be transport")
+	}
+	if IsTransportRoute(RouteFlood) {
+		t.Error("RouteFlood should not be transport")
+	}
+	if IsTransportRoute(RouteDirect) {
+		t.Error("RouteDirect should not be transport")
+	}
+}
+
+func TestPathBytesAreHops(t *testing.T) {
+	if PathBytesAreHops(PayloadTRACE) {
+		t.Error("PathBytesAreHops(PayloadTRACE) should be false")
+	}
+	// All other known payload types should return true.
+	otherTypes := []byte{0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07, 0x08, 0x0A, 0x0B, 0x0C, 0x0D, 0x0E, 0x0F}
+	for _, pt := range otherTypes {
+		if !PathBytesAreHops(pt) {
+			t.Errorf("PathBytesAreHops(0x%02X) should be true", pt)
+		}
+	}
+}

public/app.js

@@ -14,6 +14,71 @@ function isTransportRoute(rt) { return rt === 0 || rt === 3; }
 function getPathLenOffset(routeType) { return isTransportRoute(routeType) ? 5 : 1; }
 function transportBadge(rt) { return isTransportRoute(rt) ? ' <span class="badge badge-transport" title="' + routeTypeName(rt) + '">T</span>' : ''; }
 
+/**
+ * Compute breakdown byte ranges from raw_hex on the client.
+ * Mirrors cmd/server/decoder.go BuildBreakdown(). Used so per-observation raw_hex
+ * (which can differ in path length from the top-level packet) gets accurate
+ * highlighted byte ranges, instead of using the server-supplied breakdown
+ * computed once from the top-level raw_hex.
+ */
+function computeBreakdownRanges(hexString, routeType, payloadType) {
+  if (!hexString) return [];
+  const clean = hexString.replace(/\s+/g, '');
+  const bytes = clean.length / 2;
+  if (bytes < 2) return [];
+  const ranges = [];
+  // Header
+  ranges.push({ start: 0, end: 0, label: 'Header' });
+  let offset = 1;
+  if (isTransportRoute(routeType)) {
+    if (bytes < offset + 4) return ranges;
+    ranges.push({ start: offset, end: offset + 3, label: 'Transport Codes' });
+    offset += 4;
+  }
+  if (offset >= bytes) return ranges;
+  // Path Length byte
+  ranges.push({ start: offset, end: offset, label: 'Path Length' });
+  const pathByte = parseInt(clean.slice(offset * 2, offset * 2 + 2), 16);
+  offset += 1;
+  if (isNaN(pathByte)) return ranges;
+  const hashSize = (pathByte >> 6) + 1;
+  const hashCount = pathByte & 0x3F;
+  const pathBytes = hashSize * hashCount;
+  if (hashCount > 0 && offset + pathBytes <= bytes) {
+    ranges.push({ start: offset, end: offset + pathBytes - 1, label: 'Path' });
+  }
+  offset += pathBytes;
+  if (offset >= bytes) return ranges;
+  const payloadStart = offset;
+  // ADVERT (payload_type 4) gets sub-fields when full record present
+  if (payloadType === 4 && bytes - payloadStart >= 100) {
+    ranges.push({ start: payloadStart,      end: payloadStart + 31, label: 'PubKey' });
+    ranges.push({ start: payloadStart + 32, end: payloadStart + 35, label: 'Timestamp' });
+    ranges.push({ start: payloadStart + 36, end: payloadStart + 99, label: 'Signature' });
+    const appStart = payloadStart + 100;
+    if (appStart < bytes) {
+      ranges.push({ start: appStart, end: appStart, label: 'Flags' });
+      const appFlags = parseInt(clean.slice(appStart * 2, appStart * 2 + 2), 16);
+      let fOff = appStart + 1;
+      if (!isNaN(appFlags)) {
+        if ((appFlags & 0x10) && fOff + 8 <= bytes) {
+          ranges.push({ start: fOff,     end: fOff + 3, label: 'Latitude' });
+          ranges.push({ start: fOff + 4, end: fOff + 7, label: 'Longitude' });
+          fOff += 8;
+        }
+        if ((appFlags & 0x20) && fOff + 2 <= bytes) fOff += 2;
+        if ((appFlags & 0x40) && fOff + 2 <= bytes) fOff += 2;
+        if ((appFlags & 0x80) && fOff < bytes) {
+          ranges.push({ start: fOff, end: bytes - 1, label: 'Name' });
+        }
+      }
+    }
+  } else {
+    ranges.push({ start: payloadStart, end: bytes - 1, label: 'Payload' });
+  }
+  return ranges;
+}
+
 // --- Utilities ---
 const _apiPerf = { calls: 0, totalMs: 0, log: [], cacheHits: 0 };
 const _apiCache = new Map();

public/channels.js

@@ -393,17 +393,25 @@
     }
   }
 
-  // Merge user-stored keys into the channel list
+  // Merge user-stored keys into the channel list.
+  // If a stored key matches a server-known channel, mark that channel as
+  // userAdded so the ✕ button appears — otherwise the user has no way to
+  // remove a key they added but that the server already knows about.
   function mergeUserChannels() {
     var keys = ChannelDecrypt.getStoredKeys();
     var names = Object.keys(keys);
     for (var i = 0; i < names.length; i++) {
       var name = names[i];
-      // Check if channel already exists by name
-      var exists = channels.some(function (ch) {
-        return ch.name === name || ch.hash === name || ch.hash === ('user:' + name);
-      });
-      if (!exists) {
+      var matched = false;
+      for (var j = 0; j < channels.length; j++) {
+        var ch = channels[j];
+        if (ch.name === name || ch.hash === name || ch.hash === ('user:' + name)) {
+          ch.userAdded = true;
+          matched = true;
+          break;
+        }
+      }
+      if (!matched) {
         channels.push({
           hash: 'user:' + name,
           name: name,
@@ -749,19 +757,38 @@
         e.stopPropagation();
         var channelHash = removeBtn.getAttribute('data-remove-channel');
         if (!channelHash) return;
-        var chName = channelHash.startsWith('user:') ? channelHash.substring(5) : channelHash;
+        // The localStorage key is the channel name. For user:-prefixed entries
+        // strip the prefix; for server-known channels look up the channel
+        // object so we use its display name (the hash itself isn't the key).
+        var ch = channels.find(function (c) { return c.hash === channelHash; });
+        var chName = channelHash.startsWith('user:')
+          ? channelHash.substring(5)
+          : (ch && ch.name) || channelHash;
         if (!confirm('Remove channel "' + chName + '"? This will clear saved keys and cached messages.')) return;
         ChannelDecrypt.removeKey(chName);
-        // Remove from channels array
-        channels = channels.filter(function (c) { return c.hash !== channelHash; });
-        if (selectedHash === channelHash) {
-          selectedHash = null;
-          messages = [];
-          history.replaceState(null, '', '#/channels');
-          var msgEl2 = document.getElementById('chMessages');
-          if (msgEl2) msgEl2.innerHTML = '<div class="ch-empty">Choose a channel from the sidebar to view messages</div>';
-          var header2 = document.getElementById('chHeader');
-          if (header2) header2.querySelector('.ch-header-text').textContent = 'Select a channel';
+        if (channelHash.startsWith('user:')) {
+          // Pure user-added channel — drop from the list entirely.
+          channels = channels.filter(function (c) { return c.hash !== channelHash; });
+          if (selectedHash === channelHash) {
+            selectedHash = null;
+            messages = [];
+            history.replaceState(null, '', '#/channels');
+            var msgEl2 = document.getElementById('chMessages');
+            if (msgEl2) msgEl2.innerHTML = '<div class="ch-empty">Choose a channel from the sidebar to view messages</div>';
+            var header2 = document.getElementById('chHeader');
+            if (header2) header2.querySelector('.ch-header-text').textContent = 'Select a channel';
+          }
+        } else if (ch) {
+          // Server-known channel: keep the row, just unmark as user-added so
+          // the ✕ disappears until they re-add a key.
+          ch.userAdded = false;
+          // If this was the selected channel, clear decrypted messages since
+          // the key is gone — they can't be re-decrypted without re-adding it.
+          if (selectedHash === channelHash) {
+            messages = [];
+            var msgEl2 = document.getElementById('chMessages');
+            if (msgEl2) msgEl2.innerHTML = '<div class="ch-empty">Key removed — add a key to decrypt messages</div>';
+          }
         }
         renderChannelList();
         return;

public/hop-resolver.js

@@ -72,33 +72,89 @@ window.HopResolver = (function() {
   }
 
   /**
-   * Pick the best candidate using affinity first, then geo-distance fallback.
+   * Pick the best candidate by scoring against BOTH prev and next resolved hops.
+   *
+   * Strategy (in priority order):
+   * 1. Neighbor-graph edge weight: sum of edge scores to prevPubkey + nextPubkey. Pick max.
+   * 2. Geographic centroid: if no candidate has graph edges, compute centroid of
+   *    prev+next positions and pick closest candidate by haversine distance.
+   * 3. Single-anchor geo fallback: if only one neighbor is resolved, use it as anchor.
+   * 4. Original heuristic: first candidate (when no context at all).
+   *
    * @param {Array} candidates - candidates with lat/lon/pubkey/name
-   * @param {string|null} adjacentPubkey - pubkey of the previously/next resolved hop
-   * @param {Object|null} anchor - {lat, lon} for geo fallback
-   * @param {number|null} fallbackLat - fallback anchor lat (e.g. observer)
-   * @param {number|null} fallbackLon - fallback anchor lon
+   * @param {string|null} prevPubkey - pubkey of previous resolved hop
+   * @param {string|null} nextPubkey - pubkey of next resolved hop
+   * @param {Object|null} prevPos - {lat, lon} of previous resolved hop or origin
+   * @param {Object|null} nextPos - {lat, lon} of next resolved hop or observer
    * @returns {Object} best candidate
    */
-  function pickByAffinity(candidates, adjacentPubkey, anchor, fallbackLat, fallbackLon) {
-    // If we have affinity data and an adjacent hop, prefer neighbors
-    if (adjacentPubkey && Object.keys(affinityMap).length > 0) {
-      const withAffinity = candidates
-        .map(c => ({ ...c, affinity: getAffinity(adjacentPubkey, c.pubkey) }))
-        .filter(c => c.affinity > 0);
-      if (withAffinity.length > 0) {
-        withAffinity.sort((a, b) => b.affinity - a.affinity);
-        return withAffinity[0];
+  function pickByAffinity(candidates, prevPubkey, nextPubkey, prevPos, nextPos) {
+    const hasGraph = Object.keys(affinityMap).length > 0;
+    const hasAdj = prevPubkey || nextPubkey;
+
+    // Strategy 1: neighbor-graph edge weights (sum of prev + next)
+    if (hasGraph && hasAdj) {
+      const scored = candidates.map(function(c) {
+        let s = 0;
+        if (prevPubkey) s += getAffinity(prevPubkey, c.pubkey);
+        if (nextPubkey) s += getAffinity(nextPubkey, c.pubkey);
+        return { candidate: c, edgeScore: s };
+      });
+      const withEdges = scored.filter(function(s) { return s.edgeScore > 0; });
+      if (withEdges.length > 0) {
+        withEdges.sort(function(a, b) { return b.edgeScore - a.edgeScore; });
+        _traceMultiCandidate(candidates, scored, withEdges[0].candidate, 'graph');
+        return withEdges[0].candidate;
       }
     }
-    // Fallback: geo-distance sort (existing behavior)
-    const effectiveAnchor = anchor || (fallbackLat != null ? { lat: fallbackLat, lon: fallbackLon } : null);
-    if (effectiveAnchor) {
-      candidates.sort((a, b) => dist(a.lat, a.lon, effectiveAnchor.lat, effectiveAnchor.lon) - dist(b.lat, b.lon, effectiveAnchor.lat, effectiveAnchor.lon));
+
+    // Strategy 2/3: geographic — centroid of prev+next, or single anchor
+    let anchorLat = null, anchorLon = null, anchorCount = 0;
+    if (prevPos && prevPos.lat != null && prevPos.lon != null) {
+      anchorLat = (anchorLat || 0) + prevPos.lat;
+      anchorLon = (anchorLon || 0) + prevPos.lon;
+      anchorCount++;
     }
+    if (nextPos && nextPos.lat != null && nextPos.lon != null) {
+      anchorLat = (anchorLat || 0) + nextPos.lat;
+      anchorLon = (anchorLon || 0) + nextPos.lon;
+      anchorCount++;
+    }
+    if (anchorCount > 0) {
+      anchorLat /= anchorCount;
+      anchorLon /= anchorCount;
+      const geoScored = candidates.map(function(c) {
+        const d = (c.lat != null && c.lon != null && !(c.lat === 0 && c.lon === 0))
+          ? haversineKm(c.lat, c.lon, anchorLat, anchorLon) : 999999;
+        return { candidate: c, distKm: d };
+      });
+      geoScored.sort(function(a, b) { return a.distKm - b.distKm; });
+      _traceMultiCandidate(candidates, geoScored, geoScored[0].candidate, 'centroid');
+      return geoScored[0].candidate;
+    }
+
+    // Strategy 4: no context — return first candidate
+    _traceMultiCandidate(candidates, null, candidates[0], 'fallback');
     return candidates[0];
   }
 
+  /** Dev-mode console trace for multi-candidate picks */
+  function _traceMultiCandidate(candidates, scored, chosen, method) {
+    if (typeof console === 'undefined' || !console.debug) return;
+    if (candidates.length < 2) return;
+    try {
+      const prefix = candidates[0].pubkey ? candidates[0].pubkey.slice(0, 2) : '??';
+      const scoreSummary = scored ? scored.map(function(s) {
+        const pk = (s.candidate || s).pubkey || '?';
+        const val = s.edgeScore != null ? s.edgeScore : (s.distKm != null ? s.distKm + 'km' : '?');
+        return pk.slice(0, 8) + ':' + val;
+      }) : [];
+      console.debug('[hop-resolver] hash=' + prefix + ' candidates=' + candidates.length +
+        ' scored=[' + scoreSummary.join(',') + '] chose=' + (chosen.pubkey || '?').slice(0, 8) +
+        ' method=' + method);
+    } catch(e) { /* trace is best-effort */ }
+  }
+
   /**
    * Resolve an array of hex hop prefixes to node info.
    * Returns a map: { hop: {name, pubkey, lat, lon, ambiguous, unreliable} }
@@ -169,52 +225,54 @@ window.HopResolver = (function() {
       }
     }
 
-    // Forward pass
-    let lastPos = (originLat != null && originLon != null) ? { lat: originLat, lon: originLon } : null;
-    let lastResolvedPubkey = null;
-    for (let i = 0; i < hops.length; i++) {
-      const hop = hops[i];
-      if (hopPositions[hop]) {
-        lastPos = hopPositions[hop];
-        lastResolvedPubkey = resolved[hop] ? resolved[hop].pubkey : null;
-        continue;
+    // Combined disambiguation: resolve ambiguous hops using both neighbors.
+    // We iterate until no more hops can be resolved (handles cascading dependencies).
+    const originPos = (originLat != null && originLon != null) ? { lat: originLat, lon: originLon } : null;
+    const observerPos = (observerLat != null && observerLon != null) ? { lat: observerLat, lon: observerLon } : null;
+
+    let changed = true;
+    let maxIter = hops.length + 1; // prevent infinite loops
+    while (changed && maxIter-- > 0) {
+      changed = false;
+      for (let i = 0; i < hops.length; i++) {
+        const hop = hops[i];
+        if (hopPositions[hop]) continue; // already resolved
+        const r = resolved[hop];
+        if (!r || !r.ambiguous) continue;
+        const withLoc = r.candidates.filter(c => c.lat != null && c.lon != null && !(c.lat === 0 && c.lon === 0));
+        if (!withLoc.length) continue;
+
+        // Find prev resolved neighbor
+        let prevPubkey = null, prevPos = null;
+        for (let j = i - 1; j >= 0; j--) {
+          if (hopPositions[hops[j]]) {
+            prevPos = hopPositions[hops[j]];
+            prevPubkey = resolved[hops[j]] ? resolved[hops[j]].pubkey : null;
+            break;
+          }
+        }
+        if (!prevPos && originPos) prevPos = originPos;
+
+        // Find next resolved neighbor
+        let nextPubkey = null, nextPos = null;
+        for (let j = i + 1; j < hops.length; j++) {
+          if (hopPositions[hops[j]]) {
+            nextPos = hopPositions[hops[j]];
+            nextPubkey = resolved[hops[j]] ? resolved[hops[j]].pubkey : null;
+            break;
+          }
+        }
+        if (!nextPos && observerPos) nextPos = observerPos;
+
+        // Skip if we have zero context (wait for a later iteration or neighbor resolution)
+        if (!prevPubkey && !nextPubkey && !prevPos && !nextPos) continue;
+
+        const picked = pickByAffinity(withLoc, prevPubkey, nextPubkey, prevPos, nextPos);
+        r.name = picked.name;
+        r.pubkey = picked.pubkey;
+        hopPositions[hop] = { lat: picked.lat, lon: picked.lon };
+        changed = true;
       }
-      const r = resolved[hop];
-      if (!r || !r.ambiguous) continue;
-      const withLoc = r.candidates.filter(c => c.lat && c.lon && !(c.lat === 0 && c.lon === 0));
-      if (!withLoc.length) continue;
-
-      // Affinity-aware: prefer candidates that are neighbors of the previous hop
-      const picked = pickByAffinity(withLoc, lastResolvedPubkey, lastPos, i === hops.length - 1 ? observerLat : null, i === hops.length - 1 ? observerLon : null);
-      r.name = picked.name;
-      r.pubkey = picked.pubkey;
-      hopPositions[hop] = { lat: picked.lat, lon: picked.lon };
-      lastPos = hopPositions[hop];
-      lastResolvedPubkey = picked.pubkey;
-    }
-
-    // Backward pass
-    let nextPos = (observerLat != null && observerLon != null) ? { lat: observerLat, lon: observerLon } : null;
-    let nextResolvedPubkey = null;
-    for (let i = hops.length - 1; i >= 0; i--) {
-      const hop = hops[i];
-      if (hopPositions[hop]) {
-        nextPos = hopPositions[hop];
-        nextResolvedPubkey = resolved[hop] ? resolved[hop].pubkey : null;
-        continue;
-      }
-      const r = resolved[hop];
-      if (!r || !r.ambiguous) continue;
-      const withLoc = r.candidates.filter(c => c.lat && c.lon && !(c.lat === 0 && c.lon === 0));
-      if (!withLoc.length || !nextPos) continue;
-
-      // Affinity-aware: prefer candidates that are neighbors of the next hop
-      const picked = pickByAffinity(withLoc, nextResolvedPubkey, nextPos, null, null);
-      r.name = picked.name;
-      r.pubkey = picked.pubkey;
-      hopPositions[hop] = { lat: picked.lat, lon: picked.lon };
-      nextPos = hopPositions[hop];
-      nextResolvedPubkey = picked.pubkey;
     }
 
     // Sanity check: drop hops impossibly far from neighbors
@@ -276,13 +334,13 @@ window.HopResolver = (function() {
    */
   function resolveFromServer(hops, resolvedPath) {
     if (!hops || !resolvedPath || hops.length !== resolvedPath.length) return {};
-    var result = {};
-    for (var i = 0; i < hops.length; i++) {
-      var hop = hops[i];
-      var pubkey = resolvedPath[i];
+    const result = {};
+    for (let i = 0; i < hops.length; i++) {
+      const hop = hops[i];
+      const pubkey = resolvedPath[i];
       if (!pubkey) continue; // null = unresolved, leave for client-side fallback
       // O(1) lookup via pubkeyIdx built during init()
-      var node = pubkeyIdx[pubkey.toLowerCase()] || null;
+      const node = pubkeyIdx[pubkey.toLowerCase()] || null;
       result[hop] = {
         name: node ? node.name : pubkey.slice(0, 8),
         pubkey: pubkey,

public/nodes.js

@@ -1144,6 +1144,19 @@
     makeColumnsResizable('#nodesTable', 'meshcore-nodes-col-widths');
   }
 
+  /**
+   * Navigate to the full-screen node view for `pubkey` from anywhere within
+   * the nodes module. Single source of navigation truth — works regardless
+   * of current hash state (hash assignment alone is a no-op when the hash
+   * is already the target).
+   */
+  function navigateToNode(pubkey) {
+    destroy();
+    var appEl = document.getElementById('app');
+    history.replaceState(null, '', '#/nodes/' + encodeURIComponent(pubkey));
+    init(appEl, pubkey);
+  }
+
   async function selectNode(pubkey) {
     // On mobile, navigate to full-screen node view
     if (window.innerWidth <= 640) {
@@ -1307,12 +1320,11 @@
       } catch {}
     }
 
-    // #856: Wire "Details" button to navigate to full-screen node view
+    // Wire "Details" button via the unified navigateToNode helper
     var detailBtn = panel.querySelector('.node-detail-btn');
     if (detailBtn) {
       detailBtn.addEventListener('click', function() {
-        var pk = detailBtn.getAttribute('data-pubkey');
-        location.hash = '#/nodes/' + pk;
+        navigateToNode(decodeURIComponent(detailBtn.getAttribute('data-pubkey')));
       });
     }
 

public/observer-detail.js

@@ -150,6 +150,14 @@
           <div class="stat-label">First Seen</div>
           <div class="stat-value" style="font-size:0.85em">${obs.first_seen ? new Date(obs.first_seen).toLocaleDateString() : '—'}</div>
         </div>
+        <div class="stat-card">
+          <div class="stat-label">Last Status Update</div>
+          <div class="stat-value" style="font-size:0.85em">${obs.last_seen ? timeAgo(obs.last_seen) + '<br><span style="font-size:0.8em;color:var(--text-muted)">' + new Date(obs.last_seen).toLocaleString() + '</span>' : '—'}</div>
+        </div>
+        <div class="stat-card">
+          <div class="stat-label">Last Packet Observation</div>
+          <div class="stat-value" style="font-size:0.85em">${obs.last_packet_at ? timeAgo(obs.last_packet_at) + '<br><span style="font-size:0.8em;color:var(--text-muted)">' + new Date(obs.last_packet_at).toLocaleString() + '</span>' : '<span style="color:var(--text-muted)">never</span>'}</div>
+        </div>
       </div>
       <div class="mono" style="font-size:0.75em;color:var(--text-muted);margin-bottom:20px;word-break:break-all">
         ID: ${obs.id}

public/observers.js

@@ -75,6 +75,17 @@
     return { cls: 'health-red', label: 'Offline' };
   }
 
+  function packetBadge(o) {
+    if (!o.last_packet_at) return '<span title="No packets ever observed">📡⚠ never</span>';
+    const pktAgo = Date.now() - new Date(o.last_packet_at).getTime();
+    const statusAgo = o.last_seen ? Date.now() - new Date(o.last_seen).getTime() : Infinity;
+    const gap = pktAgo - statusAgo;
+    if (gap > 600000) {
+      return `<span title="Last packet ${timeAgo(o.last_packet_at)} — status is newer by ${Math.round(gap/60000)}min. Observer may be alive but not forwarding packets.">📡⚠ ${timeAgo(o.last_packet_at)}</span>`;
+    }
+    return timeAgo(o.last_packet_at);
+  }
+
   function uptimeStr(firstSeen) {
     if (!firstSeen) return '—';
     const ms = Date.now() - new Date(firstSeen).getTime();
@@ -123,7 +134,7 @@
       <div class="obs-table-scroll"><table class="data-table obs-table" id="obsTable">
         <caption class="sr-only">Observer status and statistics</caption>
         <thead><tr>
-          <th scope="col">Status</th><th scope="col">Name</th><th scope="col">Region</th><th scope="col">Last Seen</th>
+          <th scope="col">Status</th><th scope="col">Name</th><th scope="col">Region</th><th scope="col">Last Status</th><th scope="col">Last Packet</th>
           <th scope="col">Packets</th><th scope="col">Packets/Hour</th><th scope="col">Uptime</th>
         </tr></thead>
         <tbody>${filtered.map(o => {
@@ -134,6 +145,7 @@
             <td class="mono">${o.name || o.id}</td>
             <td>${o.iata ? `<span class="badge-region">${o.iata}</span>` : '—'}</td>
             <td>${timeAgo(o.last_seen)}</td>
+            <td>${packetBadge(o)}</td>
             <td>${(o.packet_count || 0).toLocaleString()}</td>
             <td>${sparkBar(o.packetsLastHour || 0, maxPktsHr)}</td>
             <td>${uptimeStr(o.first_seen)}</td>

public/packets.js

@@ -389,7 +389,7 @@
               expandedHashes.add(h);
               const obsPacket = {...data.packet, observer_id: obs.observer_id, observer_name: obs.observer_name, snr: obs.snr, rssi: obs.rssi, path_json: obs.path_json, resolved_path: obs.resolved_path, direction: obs.direction, timestamp: obs.timestamp, first_seen: obs.timestamp};
               clearParsedCache(obsPacket);
-              selectPacket(obs.id, h, {packet: obsPacket, breakdown: data.breakdown, observations: data.observations}, obs.id);
+              selectPacket(obs.id, h, {packet: obsPacket, observations: data.observations}, obs.id);
             } else {
               selectPacket(data.packet.id, h, data);
             }
@@ -519,7 +519,7 @@
               if (p.decoded_json) existing.decoded_json = p.decoded_json;
               // Update expanded children if this group is expanded
               if (expandedHashes.has(h) && existing._children) {
-                existing._children.unshift(p);
+                existing._children.unshift(clearParsedCache({...p, _isObservation: true}));
                 if (existing._children.length > 200) existing._children.length = 200;
                 sortGroupChildren(existing);
                 // Invalidate row counts — child count changed, so virtual scroll
@@ -683,10 +683,14 @@
       // Restore expanded group children (parallel fetch, Map lookup)
       if (groupByHash && expandedHashes.size > 0) {
         const expandedArr = [...expandedHashes];
+        // Fetch the full packet detail (which includes per-observation rows) for each expanded hash.
+        // Previously this used `/packets?hash=X&limit=20` which returned ONE aggregate row, causing
+        // every "child" row in the table to carry the parent packet.id instead of unique observation
+        // ids — so clicking any child pointed the side pane at the same aggregate. See #866.
         const results = await Promise.all(expandedArr.map(hash => {
           const group = hashIndex.get(hash);
           if (!group) return { hash, group: null, data: null };
-          return api(`/packets?hash=${hash}&limit=20`)
+          return api(`/packets/${hash}`)
             .then(data => ({ hash, group, data }))
             .catch(() => ({ hash, group, data: null }));
         }));
@@ -694,7 +698,15 @@
           if (!group) {
             expandedHashes.delete(hash);
           } else if (data) {
-            group._children = data.packets || [];
+            const pkt = data.packet || group;
+            // Build per-observation children. Spread (pkt, obs) so obs-level fields
+            // (id, observer_id/name, path_json, snr/rssi, timestamp, raw_hex) override
+            // the aggregate. Each child's `id` is the observation id (unique per observer).
+            const obs = data.observations || [];
+            group._children = obs.length
+              ? obs.map(o => clearParsedCache({...pkt, ...o, _isObservation: true}))
+              : [pkt];
+            group._fetchedData = { packet: pkt, observations: obs };
             sortGroupChildren(group);
           }
         }
@@ -1248,7 +1260,7 @@
             const parentData = group._fetchedData;
             const obsPacket = parentData ? {...parentData.packet, observer_id: child.observer_id, observer_name: child.observer_name, snr: child.snr, rssi: child.rssi, path_json: child.path_json, resolved_path: child.resolved_path, direction: child.direction, timestamp: child.timestamp, first_seen: child.timestamp} : child;
             if (parentData) { clearParsedCache(obsPacket); }
-            selectPacket(child.id, parentHash, {packet: obsPacket, breakdown: parentData?.breakdown, observations: parentData?.observations}, child.id);
+            selectPacket(child.id, parentHash, {packet: obsPacket, observations: parentData?.observations}, child.id);
           }
         }
         else if (action === 'select-hash') pktSelectHash(value);
@@ -1806,8 +1818,6 @@
 
   async function renderDetail(panel, data, chosenObsId) {
     const pkt = data.packet;
-    const breakdown = data.breakdown || {};
-    const ranges = breakdown.ranges || [];
     const observations = data.observations || [];
 
     // Per-observation rendering (issue #849):
@@ -1828,6 +1838,15 @@
     const decoded = getParsedDecoded(effectivePkt) || {};
     const pathHops = getParsedPath(effectivePkt) || [];
 
+    // Compute breakdown ranges from the actually-rendered raw_hex (per-observation).
+    // Single source of truth — derived from the same bytes we display, so a
+    // post-#882 per-obs raw_hex with a different path length than the top-level
+    // packet's raw_hex still gets accurate byte highlights.
+    const obsRawHexForRanges = effectivePkt.raw_hex || pkt.raw_hex || '';
+    const ranges = obsRawHexForRanges
+      ? computeBreakdownRanges(obsRawHexForRanges, pkt.route_type, pkt.payload_type)
+      : [];
+
     // Cross-check: hop count from raw_hex path_len byte vs path_json length
     const obsRawHex = effectivePkt.raw_hex || pkt.raw_hex || '';
     let rawHopCount = null;
@@ -1838,7 +1857,7 @@
       if (!isNaN(plByte)) rawHopCount = plByte & 0x3F;
     }
     if (rawHopCount != null && pathHops.length !== rawHopCount) {
-      console.warn(`[CoreScope] Hop count inconsistency for packet ${pkt.hash}: path_json has ${pathHops.length} hops but raw_hex path_len has ${rawHopCount}. Trusting raw_hex.`);
+      console.warn(`[CoreScope] Hop count inconsistency for packet ${pkt.hash}: path_json has ${pathHops.length} hops but raw_hex path_len has ${rawHopCount}. UI shows path_json.`);
     }
 
     // Resolve sender GPS — from packet directly, or from known node in DB
@@ -1975,8 +1994,10 @@
       ? `<div class="anomaly-banner" style="background:var(--warning, #f0ad4e); color:#000; padding:8px 12px; border-radius:4px; margin-bottom:8px; font-weight:600;">⚠️ Anomaly: ${escapeHtml(decoded.anomaly)}</div>`
       : '';
 
-    // Hop count display: trust raw_hex (firmware truth) over path_json
-    const displayHopCount = rawHopCount != null ? rawHopCount : pathHops.length;
+    // Hop count display: use pathHops length (= effective observation's path_json).
+    // The raw_hex/path_json mismatch warning is logged above for diagnostics; the UI
+    // must stay self-consistent — top pill names and byte breakdown rows must agree.
+    const displayHopCount = pathHops.length;
     const obsIndicator = currentObs && observations.length > 1
       ? `<span style="font-size:0.8em;color:var(--text-muted);margin-left:6px">(observation ${observations.indexOf(currentObs) + 1} of ${observations.length})</span>`
       : '';
@@ -2181,18 +2202,19 @@
     rows += fieldRow(off, 'Path Length', '0x' + (buf.slice(off * 2, off * 2 + 2) || '??'), hashCountVal === 0 ? `hash_count=0 (direct advert)` : `hash_size=${hashSizeVal} byte${hashSizeVal !== 1 ? 's' : ''}, hash_count=${hashCountVal}`);
     off += 1;
 
-    // Path — derive hop count from path_len byte (firmware truth), not aggregated _parsedPath
+    // Path — render hops from path_json (what this observation reported).
+    // Byte offsets advance by hashSize * pathHops.length to match.
     const hashSize = isNaN(pathByte0) ? 1 : ((pathByte0 >> 6) + 1);
-    if (typeof hashCountVal === 'number' && hashCountVal > 0) {
-      rows += sectionRow('Path (' + hashCountVal + ' hops)', 'section-path');
-      for (let i = 0; i < hashCountVal; i++) {
+    if (pathHops.length > 0) {
+      rows += sectionRow('Path (' + pathHops.length + ' hops)', 'section-path');
+      for (let i = 0; i < pathHops.length; i++) {
         const hopOff = off + i * hashSize;
-        const hex = buf.slice(hopOff * 2, (hopOff + hashSize) * 2).toUpperCase();
+        const hex = String(pathHops[i] || '').toUpperCase();
         const hopHtml = HopDisplay.renderHop(hex, hopNameCache[hex]);
         const label = `Hop ${i} — ${hopHtml}`;
         rows += fieldRow(hopOff, label, hex, '');
       }
-      off += hashSize * hashCountVal;
+      off += hashSize * pathHops.length;
     }
 
     // Payload
@@ -2466,7 +2488,7 @@
       renderTableRows();
       return;
     }
-    // Single fetch — gets packet + observations + path + breakdown
+    // Single fetch — gets packet + observations + path
     try {
       const data = await api(`/packets/${hash}`);
       const pkt = data.packet;

public/style.css

@@ -1556,7 +1556,7 @@ tr[data-hops]:hover { background: rgba(59,130,246,0.1); }
 
 /* #20 — Observers table horizontal scroll on mobile */
 .obs-table-scroll { overflow-x: auto; -webkit-overflow-scrolling: touch; }
-.obs-table-scroll .obs-table { min-width: 640px; }
+.obs-table-scroll .obs-table { min-width: 720px; }
 
 /* #206 — Analytics/Compare tables scroll wrappers on mobile */
 .analytics-table-scroll { overflow-x: auto; -webkit-overflow-scrolling: touch; }

test-e2e-playwright.js

@@ -15,6 +15,11 @@ async function test(name, fn) {
     results.push({ name, pass: true });
     console.log(`  \u2705 ${name}`);
   } catch (err) {
+    if (err.skip) {
+      results.push({ name, pass: true, skipped: true });
+      console.log(`  ⏭ ${name}: ${err.message}`);
+      return;
+    }
     results.push({ name, pass: false, error: err.message });
     console.log(`  \u274c ${name}: ${err.message}`);
     console.log(`\nFail-fast: stopping after first failure.`);
@@ -1778,12 +1783,343 @@ async function run() {
     }
   });
 
+  // Test: Expanded group children have unique observation ids (#866)
+  await test('Expanded group children update detail pane per-observation', async () => {
+    await page.goto(`${BASE}/#/packets`, { waitUntil: 'domcontentloaded' });
+    // Ensure grouped mode and wide time window
+    await page.evaluate(() => {
+      localStorage.setItem('meshcore-time-window', '525600');
+      localStorage.setItem('meshcore-groupbyhash', 'true');
+    });
+    await page.reload({ waitUntil: 'load' });
+    await page.waitForSelector('table tbody tr', { timeout: 15000 });
+
+    // Find a group row with observation_count > 1 (has expand button)
+    const expandBtn = await page.$('table tbody tr .expand-btn, table tbody tr [data-expand]');
+    if (!expandBtn) {
+      console.log('    ℹ️  No expandable groups found — skipping child assertion');
+      return;
+    }
+
+    // Click expand and wait for the /packets/<hash> detail API call
+    const [detailResp] = await Promise.all([
+      page.waitForResponse(resp => {
+        const u = new URL(resp.url(), BASE);
+        // Match /api/packets/<hash> but not /api/packets?... or /api/packets/observations
+        return /\/api\/packets\/[A-Fa-f0-9]+$/.test(u.pathname) && resp.status() === 200;
+      }, { timeout: 15000 }),
+      expandBtn.click(),
+    ]);
+    assert(detailResp, 'Expected /api/packets/<hash> response on expand');
+
+    // Wait for child rows to appear
+    await page.waitForSelector('table tbody tr.child-row, table tbody tr[class*="child"]', { timeout: 5000 });
+    const childRows = await page.$$('table tbody tr.child-row, table tbody tr[class*="child"]');
+    if (childRows.length < 2) {
+      console.log('    ℹ️  Group has < 2 children — skipping per-observation assertion');
+      return;
+    }
+
+    // Click first child row
+    await childRows[0].click();
+    await page.waitForFunction(() => {
+      const panel = document.getElementById('pktRight');
+      return panel && !panel.classList.contains('empty') && panel.textContent.trim().length > 0;
+    }, { timeout: 10000 });
+    const content1 = await page.$eval('#pktRight', el => el.textContent.trim());
+    const url1 = page.url();
+
+    // Click second child row
+    await childRows[1].click();
+    await page.waitForTimeout(500);
+    const content2 = await page.$eval('#pktRight', el => el.textContent.trim());
+    const url2 = page.url();
+
+    // URL should contain ?obs= with a real observation id
+    assert(url1.includes('obs=') || url2.includes('obs='), `URL should contain obs= parameter, got: ${url1}`);
+
+    // The two children should show different detail pane content (different observers)
+    // At minimum, the URL obs= values should differ
+    if (url1.includes('obs=') && url2.includes('obs=')) {
+      const obs1 = new URL(url1).hash.match(/obs=(\d+)/)?.[1];
+      const obs2 = new URL(url2).hash.match(/obs=(\d+)/)?.[1];
+      if (obs1 && obs2) {
+        assert(obs1 !== obs2, `Two children should have different obs ids, both got obs=${obs1}`);
+      }
+    }
+
+    // Verify obs id is NOT the aggregate packet id (the bug from #866)
+    const obsMatch = url2.match(/obs=(\d+)/);
+    if (obsMatch) {
+      const detailJson = await detailResp.json().catch(() => null);
+      if (detailJson?.packet?.id) {
+        const aggId = String(detailJson.packet.id);
+        // At least one child obs id should differ from the aggregate packet id
+        const obs1 = url1.match(/obs=(\d+)/)?.[1];
+        const obs2 = url2.match(/obs=(\d+)/)?.[1];
+        const allSameAsAgg = obs1 === aggId && obs2 === aggId;
+        assert(!allSameAsAgg, `Child obs ids should not all equal aggregate packet.id (${aggId})`);
+      }
+    }
+  });
+
+  // Test: per-observation raw_hex — hex pane updates when switching observations (#881)
+  await test('Packet detail hex pane updates per observation', async () => {
+    await page.goto(BASE + '#/packets', { waitUntil: 'domcontentloaded' });
+    await page.waitForSelector('table tbody tr', { timeout: 15000 });
+    await page.waitForTimeout(500);
+
+    // Try clicking packet rows to find one with multiple observations
+    const rows = await page.$$('table tbody tr[data-action]');
+    let obsRows = [];
+    for (let i = 0; i < Math.min(rows.length, 10); i++) {
+      await rows[i].click({ timeout: 3000 }).catch(() => null);
+      await page.waitForTimeout(600);
+      obsRows = await page.$$('.detail-obs-row');
+      if (obsRows.length >= 2) break;
+    }
+
+    if (obsRows.length < 2) {
+      console.log('    ⏭ Skipped: no packet with ≥2 observations found in first 10 rows');
+      return;
+    }
+
+    // Click first observation, capture hex dump
+    await obsRows[0].click({ timeout: 5000 });
+    await page.waitForTimeout(500);
+    const hex1 = await page.$eval('.hex-dump', el => el.textContent).catch(() => '');
+
+    // Click second observation, capture hex dump
+    await obsRows[1].click({ timeout: 5000 });
+    await page.waitForTimeout(500);
+    const hex2 = await page.$eval('.hex-dump', el => el.textContent).catch(() => '');
+
+    // If both have content and differ, the feature works
+    if (hex1 && hex2 && hex1 !== hex2) {
+      console.log('    ✓ Hex pane content differs between observations');
+    } else if (hex1 && hex2 && hex1 === hex2) {
+      console.log('    ⏭ Hex same for both observations (likely historical NULL raw_hex — OK)');
+    } else {
+      console.log('    ⏭ Could not capture hex content from both observations');
+    }
+  });
+
+  // Test: path pill (top) and byte breakdown (bottom) agree on hop count
+  // Regression for visual mismatch where badge said "1 hop" but path text listed N names
+  await test('Packet detail path pill and byte breakdown agree on hop count', async () => {
+    await page.goto(BASE + '#/packets', { waitUntil: 'domcontentloaded' });
+    await page.waitForSelector('table tbody tr', { timeout: 15000 });
+    await page.waitForTimeout(500);
+
+    // Click rows until we find one whose detail pane renders a multi-hop path
+    const rows = await page.$$('table tbody tr[data-action]');
+    let found = false;
+    for (let i = 0; i < Math.min(rows.length, 15); i++) {
+      await rows[i].click({ timeout: 3000 }).catch(() => null);
+      await page.waitForTimeout(500);
+
+      const result = await page.evaluate(() => {
+        // Path pill: <dt>Path</dt><dd><span class="badge ...">N hops</span> ...names...</dd>
+        const dts = document.querySelectorAll('dl.detail-meta dt');
+        let pillBadgeCount = null;
+        let pillNameCount = null;
+        for (const dt of dts) {
+          if (dt.textContent.trim() === 'Path') {
+            const dd = dt.nextElementSibling;
+            if (!dd) break;
+            const badge = dd.querySelector('.badge');
+            if (badge) {
+              const m = badge.textContent.match(/(\d+)\s*hop/);
+              if (m) pillBadgeCount = parseInt(m[1], 10);
+            }
+            // Count rendered hop links/spans (HopDisplay.renderHop output)
+            const hops = dd.querySelectorAll('.hop-link, [data-hop-link], .hop-named, .hop-anonymous');
+            pillNameCount = hops.length;
+            break;
+          }
+        }
+        // Byte breakdown: section row "Path (N hops)" + N "Hop X — ..." rows
+        let breakdownSectionCount = null;
+        let breakdownRowCount = 0;
+        const fieldTable = document.querySelector('table.field-table');
+        if (fieldTable) {
+          for (const tr of fieldTable.querySelectorAll('tr')) {
+            const txt = tr.textContent.trim();
+            const sec = txt.match(/^Path\s*\((\d+)\s*hops?\)/);
+            if (sec) breakdownSectionCount = parseInt(sec[1], 10);
+            if (/^\s*\d+\s*Hop\s+\d+\s*—/.test(txt) || /^Hop\s+\d+\s*—/.test(txt.replace(/^\d+/, '').trim())) {
+              breakdownRowCount++;
+            }
+          }
+        }
+        return { pillBadgeCount, pillNameCount, breakdownSectionCount, breakdownRowCount };
+      });
+
+      if (result.pillBadgeCount && result.pillBadgeCount > 0 && result.breakdownSectionCount != null) {
+        found = true;
+        // Top badge count must equal bottom section count
+        assert(result.pillBadgeCount === result.breakdownSectionCount,
+          `Path pill badge says ${result.pillBadgeCount} hops but byte breakdown says ${result.breakdownSectionCount} hops`);
+        // Number of rendered hop names in pill should also match (within 1, since renderPath may add separators)
+        if (result.pillNameCount != null && result.pillNameCount > 0) {
+          assert(Math.abs(result.pillNameCount - result.pillBadgeCount) <= 1,
+            `Path pill badge ${result.pillBadgeCount} but rendered ${result.pillNameCount} hop names`);
+        }
+        // And breakdown rendered rows should match its own section count
+        assert(result.breakdownRowCount > 0,
+          'breakdown rows selector matched nothing — selector or DOM changed');
+        assert(result.breakdownRowCount === result.breakdownSectionCount,
+          `Byte breakdown section says ${result.breakdownSectionCount} hops but rendered ${result.breakdownRowCount} hop rows`);
+        console.log(`    ✓ Path pill (${result.pillBadgeCount}) and byte breakdown (${result.breakdownSectionCount}) agree`);
+        break;
+      }
+    }
+    if (!found) {
+      if (process.env.E2E_REQUIRE_PATH_TEST === '1') {
+        throw new Error('BLOCKED — no multi-hop packet found in first 15 rows (E2E_REQUIRE_PATH_TEST=1 requires it)');
+      }
+      const skipErr = new Error('SKIP: No multi-hop packet with byte breakdown found in first 15 rows — needs fixture');
+      skipErr.skip = true;
+      throw skipErr;
+    }
+  });
+
+  // Test: hex-strip color spans match the labeled byte rows (per-obs raw_hex).
+  // Regression #891: server-supplied breakdown was computed once from top-level
+  // raw_hex, so per-observation rendering had off-by-N highlights vs the labels.
+  await test('Packet detail hex strip Path range matches hop row count', async () => {
+    await page.goto(BASE + '#/packets', { waitUntil: 'domcontentloaded' });
+    await page.waitForSelector('table tbody tr', { timeout: 15000 });
+    await page.waitForTimeout(500);
+
+    const rows = await page.$$('table tbody tr[data-action]');
+    let checked = 0;
+    for (let i = 0; i < Math.min(rows.length, 25) && checked < 3; i++) {
+      await rows[i].click({ timeout: 3000 }).catch(() => null);
+      await page.waitForTimeout(400);
+
+      const result = await page.evaluate(() => {
+        const dump = document.querySelector('.hex-dump');
+        const fieldTable = document.querySelector('table.field-table');
+        if (!dump || !fieldTable) return null;
+        const pathSpan = dump.querySelector('span.hex-byte.hex-path');
+        const pathBytes = pathSpan ? pathSpan.textContent.trim().split(/\s+/).filter(Boolean).length : 0;
+        const hopRows = [];
+        for (const tr of fieldTable.querySelectorAll('tr')) {
+          const cells = [...tr.cells].map(c => c.textContent.trim());
+          if (cells.length >= 2 && /^Hop\s+\d+/.test(cells[1])) hopRows.push(cells[2]);
+        }
+        return { pathBytes, hopRows };
+      });
+
+      if (!result || (result.pathBytes === 0 && result.hopRows.length === 0)) continue;
+      checked++;
+      // Either both zero, or the count of bytes inside hex-path == hop rows.
+      // (For multi-byte hash sizes this is bytes-per-hop * hops; for hash_size=1 it's just hops.)
+      // The simpler invariant: if there are hop rows, hex-path span must exist and have at least
+      // as many bytes as there are hops (== exactly hops * hash_size).
+      assert(result.hopRows.length > 0,
+        `row ${i}: hex-path span has ${result.pathBytes} bytes but no hop rows in the labeled table`);
+      assert(result.pathBytes >= result.hopRows.length,
+        `row ${i}: hex-path has ${result.pathBytes} bytes but ${result.hopRows.length} hop rows — strip and labels disagree`);
+      assert(result.pathBytes % result.hopRows.length === 0,
+        `row ${i}: hex-path has ${result.pathBytes} bytes but ${result.hopRows.length} hop rows — bytes/hops not divisible (hash_size violated)`);
+      console.log(`    ✓ row ${i}: hex-path ${result.pathBytes} bytes / ${result.hopRows.length} hop rows (hash_size=${result.pathBytes / result.hopRows.length})`);
+    }
+    if (checked === 0) {
+      const skipErr = new Error('SKIP: no packet with rendered hex strip + hop rows found in first 25 rows');
+      skipErr.skip = true;
+      throw skipErr;
+    }
+  });
+
+  // Test: clicking a different observation row re-renders strip + breakdown consistently.
+  // Regression: observations of the same packet hash have different raw_hex (#882),
+  // so picking a different obs must recompute the byte ranges, not reuse the old ones.
+  await test('Packet detail switches consistently across observations', async () => {
+    await page.goto(BASE + '#/packets?groupByHash=1', { waitUntil: 'domcontentloaded' });
+    await page.waitForSelector('table tbody tr', { timeout: 15000 });
+    await page.waitForTimeout(500);
+
+    let opened = false;
+    const groupRows = await page.$$('table tbody tr[data-action]');
+    for (let i = 0; i < Math.min(groupRows.length, 10); i++) {
+      await groupRows[i].click({ timeout: 3000 }).catch(() => null);
+      await page.waitForTimeout(400);
+      const obsCount = await page.evaluate(() => {
+        return document.querySelectorAll('table.observations-table tbody tr, .obs-row').length;
+      });
+      if (obsCount >= 2) { opened = true; break; }
+    }
+    if (!opened) {
+      const skipErr = new Error('SKIP: no multi-observation packet found in first 10 group rows');
+      skipErr.skip = true;
+      throw skipErr;
+    }
+
+    async function snapshot() {
+      return page.evaluate(() => {
+        const dump = document.querySelector('.hex-dump');
+        const fieldTable = document.querySelector('table.field-table');
+        if (!dump || !fieldTable) return null;
+        const pathSpan = dump.querySelector('span.hex-byte.hex-path');
+        const pathBytes = pathSpan ? pathSpan.textContent.trim().split(/\s+/).filter(Boolean).length : 0;
+        const hopRows = [];
+        for (const tr of fieldTable.querySelectorAll('tr')) {
+          const cells = [...tr.cells].map(c => c.textContent.trim());
+          if (cells.length >= 2 && /^Hop\s+\d+/.test(cells[1])) hopRows.push(cells[2]);
+        }
+        const rawHexParts = [...dump.querySelectorAll('span.hex-byte')].map(s => s.textContent.trim());
+        return { pathBytes, hopCount: hopRows.length, rawHexJoined: rawHexParts.join('|') };
+      });
+    }
+
+    const snapA = await snapshot();
+    assert(snapA, 'first snapshot must have hex dump + field table');
+    assert(snapA.hopCount === 0 || snapA.pathBytes >= snapA.hopCount,
+      `obs A inconsistent: hex-path ${snapA.pathBytes} bytes vs ${snapA.hopCount} hop rows`);
+
+    const switched = await page.evaluate(() => {
+      const obsRows = [...document.querySelectorAll('table.observations-table tbody tr, .obs-row')];
+      if (obsRows.length < 2) return false;
+      obsRows[1].click();
+      return true;
+    });
+    assert(switched, 'should click second observation row');
+    await page.waitForTimeout(500);
+
+    const snapB = await snapshot();
+    assert(snapB, 'second snapshot must have hex dump + field table');
+    assert(snapB.hopCount === 0 || snapB.pathBytes >= snapB.hopCount,
+      `obs B inconsistent: hex-path ${snapB.pathBytes} bytes vs ${snapB.hopCount} hop rows`);
+    console.log(`    ✓ obs A: ${snapA.pathBytes} path bytes / ${snapA.hopCount} hops; obs B: ${snapB.pathBytes} / ${snapB.hopCount}`);
+  });
+
+  // Test: clicking the 🔍 Details button in the nodes side panel navigates to
+  // the full-screen node detail view. Regression: hash already === target,
+  // so location.hash assignment was a no-op and the panel stayed open.
+  await test('Nodes side panel Details button opens full-screen view', async () => {
+    await page.goto(BASE + '#/nodes', { waitUntil: 'domcontentloaded' });
+    await page.waitForSelector('table tbody tr[data-action]', { timeout: 15000 });
+    await page.waitForTimeout(500);
+    // Open side panel
+    await page.click('table tbody tr[data-action]');
+    await page.waitForSelector('#nodesRight .node-detail-btn', { timeout: 5000 });
+    // Click Details
+    await page.click('#nodesRight .node-detail-btn');
+    // Wait for full-screen view to appear
+    await page.waitForSelector('.node-fullscreen', { timeout: 5000 });
+    const isFullScreen = await page.evaluate(() => !!document.querySelector('.node-fullscreen'));
+    assert(isFullScreen, 'Details button should open full-screen node view');
+  });
+
   await browser.close();
 
   // Summary
-  const passed = results.filter(r => r.pass).length;
+  const skipped = results.filter(r => r.skipped).length;
+  const passed = results.filter(r => r.pass && !r.skipped).length;
   const failed = results.filter(r => !r.pass).length;
-  console.log(`\n${passed}/${results.length} tests passed${failed ? `, ${failed} failed` : ''}`);
+  console.log(`\n${passed}/${results.length} tests passed${skipped ? `, ${skipped} skipped` : ''}${failed ? `, ${failed} failed` : ''}`);
   process.exit(failed > 0 ? 1 : 0);
 }
 

test-frontend-helpers.js

@@ -690,6 +690,88 @@ console.log('\n=== haversineKm (hop-resolver.js) ===');
   });
 }
 
+// ===== pickByAffinity — neighbor-graph + centroid scoring (#874) =====
+console.log('\n=== pickByAffinity neighbor-graph scoring (#874) ===');
+{
+  const ctx = makeSandbox();
+  ctx.IATA_COORDS_GEO = {};
+  loadInCtx(ctx, 'public/hop-resolver.js');
+  const HR = ctx.window.HopResolver;
+
+  // Two nodes sharing prefix "ab", hundreds of km apart.
+  // NodeSF is near San Francisco, NodeDEN is near Denver.
+  const nodeSF = { public_key: 'ab11111111111111', name: 'NodeSF', lat: 37.7, lon: -122.4 };
+  const nodeDEN = { public_key: 'ab22222222222222', name: 'NodeDEN', lat: 39.7, lon: -104.9 };
+  // A known neighbor of NodeSF (in the graph)
+  const nodeNeighbor = { public_key: 'cc33333333333333', name: 'SFNeighbor', lat: 37.8, lon: -122.3 };
+  // Another known node near Denver
+  const nodeDenNeighbor = { public_key: 'dd44444444444444', name: 'DENNeighbor', lat: 39.8, lon: -105.0 };
+
+  test('#874: graph edge scoring picks correct regional candidate (SF)', () => {
+    HR.init([nodeSF, nodeDEN, nodeNeighbor, nodeDenNeighbor]);
+    HR.setAffinity({ edges: [
+      { source: 'cc33333333333333', target: 'ab11111111111111', weight: 5 },
+      { source: 'dd44444444444444', target: 'ab22222222222222', weight: 5 },
+    ]});
+    // Path: SFNeighbor → [ab??] → DENNeighbor
+    // With graph edges, ab11 (NodeSF) has edge to SFNeighbor, ab22 (NodeDEN) has edge to DENNeighbor
+    // Prev=SFNeighbor, Next=DENNeighbor → both have score 5, but SFNeighbor edge only to ab11
+    const result = HR.resolve(['cc', 'ab', 'dd'],
+      null, null, null, null);
+    assert.strictEqual(result['ab'].name, 'NodeSF',
+      'Should pick NodeSF because it has a graph edge to prev hop SFNeighbor');
+  });
+
+  test('#874: graph edge scoring — next hop breaks tie', () => {
+    HR.init([nodeSF, nodeDEN, nodeNeighbor, nodeDenNeighbor]);
+    HR.setAffinity({ edges: [
+      { source: 'dd44444444444444', target: 'ab22222222222222', weight: 8 },
+      // No edge from SFNeighbor to either ab node
+    ]});
+    // Path: SFNeighbor → [ab??] → DENNeighbor
+    // Only ab22 (NodeDEN) has edge to DENNeighbor (next hop)
+    const result = HR.resolve(['cc', 'ab', 'dd'],
+      null, null, null, null);
+    assert.strictEqual(result['ab'].name, 'NodeDEN',
+      'Should pick NodeDEN because it has graph edge to next hop DENNeighbor');
+  });
+
+  test('#874: centroid fallback when no graph edges exist', () => {
+    HR.init([nodeSF, nodeDEN, nodeNeighbor]);
+    HR.setAffinity({ edges: [] }); // no edges at all
+    // Path: SFNeighbor → [ab??]
+    // SFNeighbor is at (37.8, -122.3), centroid is just that point
+    // NodeSF (37.7, -122.4) is ~14km away, NodeDEN (39.7, -104.9) is ~1500km away
+    const result = HR.resolve(['cc', 'ab'],
+      null, null, null, null);
+    assert.strictEqual(result['ab'].name, 'NodeSF',
+      'Should pick NodeSF via centroid proximity to SFNeighbor');
+  });
+
+  test('#874: centroid uses average of prev+next positions', () => {
+    // Prev near SF, next near Denver → centroid is midpoint (~Nevada)
+    // NodeDEN is closer to Nevada midpoint than NodeSF
+    const nodeMid = { public_key: 'ee55555555555555', name: 'MidNode', lat: 38.5, lon: -114.0 };
+    HR.init([nodeSF, nodeDEN, nodeNeighbor, nodeDenNeighbor, nodeMid]);
+    HR.setAffinity({ edges: [] });
+    // Path: SFNeighbor → [ab??] → DENNeighbor
+    // centroid = avg(37.8,-122.3, 39.8,-105.0) = (38.8, -113.65) — closer to Denver
+    const result = HR.resolve(['cc', 'ab', 'dd'],
+      null, null, null, null);
+    assert.strictEqual(result['ab'].name, 'NodeDEN',
+      'Should pick NodeDEN because centroid of SF+Denver neighbors is closer to Denver');
+  });
+
+  test('#874: fallback when no context at all', () => {
+    HR.init([nodeSF, nodeDEN]);
+    HR.setAffinity({ edges: [] });
+    // Single ambiguous hop, no origin/observer, no neighbors
+    const result = HR.resolve(['ab'], null, null, null, null);
+    assert.ok(result['ab'].ambiguous || result['ab'].name != null,
+      'Should resolve to some candidate without crashing');
+  });
+}
+
 // ===== SNR/RSSI Number casting =====
 {
   // These test the pattern used in observer-detail.js, home.js, traces.js, live.js
@@ -1722,6 +1804,128 @@ console.log('\n=== app.js: formatEngineBadge ===');
   });
 }
 
+// ===== APP.JS: computeBreakdownRanges =====
+console.log('\n=== app.js: computeBreakdownRanges ===');
+{
+  const ctx = makeSandbox();
+  loadInCtx(ctx, 'public/roles.js');
+  loadInCtx(ctx, 'public/app.js');
+  const computeBreakdownRanges = ctx.computeBreakdownRanges;
+
+  function findRange(ranges, label) {
+    return ranges.find(r => r.label === label);
+  }
+
+  test('returns [] for empty hex', () => {
+    assert.deepEqual(computeBreakdownRanges('', 1, 5), []);
+  });
+
+  test('returns [] for too-short hex (< 2 bytes)', () => {
+    assert.deepEqual(computeBreakdownRanges('15', 1, 5), []);
+  });
+
+  test('FLOOD non-transport: 4-hop hash_size=1', () => {
+    // header=15, plb=04 → hash_size=1, hash_count=4
+    // bytes: 15 04 90 FA F9 10 6E 01 D9
+    const r = computeBreakdownRanges('150490FAF910 6E01D9'.replace(/\s/g,''), 1, 5);
+    assert.deepEqual(findRange(r, 'Header'),      { start: 0, end: 0, label: 'Header' });
+    assert.deepEqual(findRange(r, 'Path Length'), { start: 1, end: 1, label: 'Path Length' });
+    assert.deepEqual(findRange(r, 'Path'),        { start: 2, end: 5, label: 'Path' });
+    assert.deepEqual(findRange(r, 'Payload'),     { start: 6, end: 8, label: 'Payload' });
+    assert.strictEqual(findRange(r, 'Transport Codes'), undefined);
+  });
+
+  test('FLOOD non-transport: 7-hop hash_size=1', () => {
+    // header=15, plb=07
+    const hex = '15077f6d7d1cadeca33988fd95e0851ebf01ea12e1879e';
+    const r = computeBreakdownRanges(hex, 1, 5);
+    assert.deepEqual(findRange(r, 'Path'), { start: 2, end: 8, label: 'Path' });
+    const payload = findRange(r, 'Payload');
+    assert.strictEqual(payload.start, 9, 'payload starts after the 7 path bytes');
+  });
+
+  test('FLOOD non-transport: 8-hop hash_size=1', () => {
+    const hex = '1508' + '11223344556677AA' + 'BBCCDD';
+    const r = computeBreakdownRanges(hex, 1, 5);
+    assert.deepEqual(findRange(r, 'Path'), { start: 2, end: 9, label: 'Path' });
+    assert.deepEqual(findRange(r, 'Payload'), { start: 10, end: 12, label: 'Payload' });
+  });
+
+  test('Direct advert: 0-hop, no Path range', () => {
+    // plb=00 → 0 hops; expect Path Length but NO Path range
+    const r = computeBreakdownRanges('1100AABBCCDD', 1, 4);
+    assert.deepEqual(findRange(r, 'Path Length'), { start: 1, end: 1, label: 'Path Length' });
+    assert.strictEqual(findRange(r, 'Path'), undefined);
+  });
+
+  test('Transport route shifts path-length offset by 4', () => {
+    // route_type=0 (TRANSPORT_FLOOD): bytes 1..4 are Transport Codes
+    // header=14, transport=AABBCCDD, plb=02, hops=11 22, payload=99
+    const hex = '14AABBCCDD021122' + '99';
+    const r = computeBreakdownRanges(hex, 0, 5);
+    assert.deepEqual(findRange(r, 'Transport Codes'), { start: 1, end: 4, label: 'Transport Codes' });
+    assert.deepEqual(findRange(r, 'Path Length'),     { start: 5, end: 5, label: 'Path Length' });
+    assert.deepEqual(findRange(r, 'Path'),            { start: 6, end: 7, label: 'Path' });
+    assert.deepEqual(findRange(r, 'Payload'),         { start: 8, end: 8, label: 'Payload' });
+  });
+
+  test('hash_size=2 (plb top bits=01): 4 hops × 2 bytes', () => {
+    // plb = 01 0001 00 = 0x44 → hash_size=2, hash_count=4 → 8 path bytes
+    const hex = '15' + '44' + 'AABB' + 'CCDD' + 'EEFF' + '1122' + '9988';
+    const r = computeBreakdownRanges(hex, 1, 5);
+    assert.deepEqual(findRange(r, 'Path'), { start: 2, end: 9, label: 'Path' });
+    assert.deepEqual(findRange(r, 'Payload'), { start: 10, end: 11, label: 'Payload' });
+  });
+
+  test('hash_size=3 (plb top bits=10): 2 hops × 3 bytes', () => {
+    // plb = 10 0000 10 = 0x82 → hash_size=3, hash_count=2 → 6 path bytes
+    const hex = '15' + '82' + 'AABBCC' + 'DDEEFF' + '99';
+    const r = computeBreakdownRanges(hex, 1, 5);
+    assert.deepEqual(findRange(r, 'Path'), { start: 2, end: 7, label: 'Path' });
+    assert.deepEqual(findRange(r, 'Payload'), { start: 8, end: 8, label: 'Payload' });
+  });
+
+  test('hash_size=4 (plb top bits=11): 2 hops × 4 bytes', () => {
+    // plb = 11 0000 10 = 0xC2 → hash_size=4, hash_count=2 → 8 path bytes
+    const hex = '15' + 'C2' + 'AABBCCDD' + 'EEFF1122' + '99887766';
+    const r = computeBreakdownRanges(hex, 1, 5);
+    assert.deepEqual(findRange(r, 'Path'), { start: 2, end: 9, label: 'Path' });
+    assert.deepEqual(findRange(r, 'Payload'), { start: 10, end: 13, label: 'Payload' });
+  });
+
+  test('truncated path: not enough bytes → no Path range', () => {
+    // plb=04 says 4 hops but only 2 bytes remain
+    const hex = '1504AABB';
+    const r = computeBreakdownRanges(hex, 1, 5);
+    assert.strictEqual(findRange(r, 'Path'), undefined);
+  });
+
+  test('ADVERT (payload_type=4) with full record: PubKey/Timestamp/Signature/Flags', () => {
+    // header=11, plb=00 (direct advert)
+    // payload: 32 bytes pubkey + 4 bytes ts + 64 bytes sig + 1 byte flags
+    const pubkey = 'AB'.repeat(32);
+    const ts = '11223344';
+    const sig = 'CD'.repeat(64);
+    const flags = '00';
+    const hex = '1100' + pubkey + ts + sig + flags;
+    const r = computeBreakdownRanges(hex, 1, 4);
+    assert.deepEqual(findRange(r, 'PubKey'),    { start: 2,  end: 33,  label: 'PubKey' });
+    assert.deepEqual(findRange(r, 'Timestamp'), { start: 34, end: 37,  label: 'Timestamp' });
+    assert.deepEqual(findRange(r, 'Signature'), { start: 38, end: 101, label: 'Signature' });
+    assert.deepEqual(findRange(r, 'Flags'),     { start: 102, end: 102, label: 'Flags' });
+  });
+
+  test('NaN-safe: malformed path-length byte produces no Path range', () => {
+    // hex with non-hex char in plb position would parseInt-fail → bail
+    // Use a 1-byte payload that makes pathByte parseInt produce NaN-ish via X
+    // (parseInt of 'XY' is NaN). Since fs reads only hex chars, simulate via short hex.
+    // Easier: empty string already returns []; 1-byte returns []. Both covered above.
+    // Use plb=FF (hash_size=4, hash_count=63) too long for input → no Path
+    const r = computeBreakdownRanges('15FF' + 'AA', 1, 5);
+    assert.strictEqual(findRange(r, 'Path'), undefined);
+  });
+}
+
 // ===== APP.JS: isTransportRoute + transportBadge =====
 console.log('\n=== app.js: isTransportRoute + transportBadge ===');
 {
@@ -5462,40 +5666,33 @@ console.log('\n=== packets.js: buildFieldTable hop count from path_len (#844) ==
   loadInCtx(ftCtx, 'public/packets.js');
   const { buildFieldTable } = ftCtx.window._packetsTestAPI;
 
-  test('#844: byte breakdown uses path_len hop count, not aggregated _parsedPath', () => {
+  test('#885: byte breakdown uses pathHops length (single source of truth)', () => {
+    // After #885 the byte breakdown agrees with the path pill: both render
+    // from the per-observation path_json. raw_hex is the underlying bytes
+    // for that same observation, so consistency is by construction.
     // path_len = 0x42 → hash_size=2, hash_count=2
     // raw_hex: header(11) + path_len(42) + hop0(41B1) + hop1(27D7) + pubkey(32 bytes)...
     const pubkey = 'C0DEDAD4'.padEnd(64, '0'); // 32 bytes = 64 hex chars
     const raw = '1142' + '41B1' + '27D7' + pubkey + '00000000' + '0'.repeat(128);
     const pkt = { raw_hex: raw, route_type: 1, payload_type: 0 };
-    // Pass aggregated pathHops with 7 hops (mismatched)
-    const pathHops = ['41B1', '5EB0', '1000', '2DD2', '52F8', '9535', '762B'];
+    // Per-obs path_json IS the source of truth — pass the 2 hops that match raw_hex.
+    const pathHops = ['41B1', '27D7'];
     const html = buildFieldTable(pkt, {}, pathHops, {});
 
-    // Section header should say "2 hops", not "7 hops"
-    assert.ok(html.includes('Path (2 hops)'), 'Should show "Path (2 hops)" from path_len, got: ' +
-      (html.match(/Path \(\d+ hops\)/)?.[0] || 'no match'));
-    assert.ok(!html.includes('Path (7 hops)'), 'Should NOT show 7 hops from aggregated path');
-
-    // Should contain hop values from raw_hex
+    assert.ok(html.includes('Path (2 hops)'), 'Should show "Path (2 hops)"');
     assert.ok(html.includes('41B1'), 'Should show hop 0 = 41B1');
     assert.ok(html.includes('27D7'), 'Should show hop 1 = 27D7');
-
-    // Should NOT contain hops from aggregated path that aren't in raw_hex
-    assert.ok(!html.includes('5EB0'), 'Should NOT show aggregated hop 5EB0');
-    assert.ok(!html.includes('9535'), 'Should NOT show aggregated hop 9535');
   });
 
-  test('#844: pubkey offset correct after 2-hop path (not after 7-hop)', () => {
+  test('#885: pubkey offset advances by hashSize * pathHops.length', () => {
     const pubkey = 'C0DEDAD4'.padEnd(64, '0');
     const raw = '1142' + '41B1' + '27D7' + pubkey + '00000000' + '0'.repeat(128);
     const pkt = { raw_hex: raw, route_type: 1, payload_type: 0 };
-    const html = buildFieldTable(pkt, { type: 'ADVERT', pubKey: pubkey }, ['41B1','5EB0','1000','2DD2','52F8','9535','762B'], {});
+    const html = buildFieldTable(pkt, { type: 'ADVERT', pubKey: pubkey }, ['41B1', '27D7'], {});
 
     // Public Key should be at offset 6 (1 header + 1 path_len + 2*2 hops = 6)
-    // Not at offset 16 (1 + 1 + 2*7 = 16)
     assert.ok(html.includes('>6<') || html.includes('"6"'),
-      'Public Key should be at offset 6, not 16');
+      'Public Key should be at offset 6');
   });
 
   test('#844: hashCountVal=0 (direct advert) skips Path section', () => {

test-hop-resolver-affinity.js

@@ -95,5 +95,27 @@ const result6 = HopResolver.resolve(['ee44'], null, null, null, null, null);
 assert(result6['ee44'].name === 'NodeD', 'Unique prefix resolves directly — got: ' + result6['ee44'].name);
 assert(!result6['ee44'].ambiguous, 'Should not be marked ambiguous');
 
+// Test 7: lat=0 / lon=0 candidates are NOT excluded (equator/prime-meridian bug fix)
+console.log('\nTest 7: lat=0 / lon=0 candidates are included in geo scoring');
+const nodeEquator = { public_key: 'ab5555', name: 'EquatorNode', lat: 0, lon: 10 };
+const nodeFar = { public_key: 'ab6666', name: 'FarNode', lat: 60, lon: 60 };
+const anchorNearEq = { public_key: 'cd7777', name: 'AnchorEq', lat: 1, lon: 11 };
+HopResolver.init([nodeEquator, nodeFar, anchorNearEq]);
+HopResolver.setAffinity({});
+// Anchor near equator — EquatorNode (0,10) should be geo-closest
+const result7 = HopResolver.resolve(['cd77', 'ab'], 1.0, 11.0, null, null, null);
+assert(result7['ab'].name === 'EquatorNode',
+  'lat=0 candidate should be included and win by geo — got: ' + result7['ab'].name);
+
+// Test 8: lon=0 candidate is also included
+console.log('\nTest 8: lon=0 candidate is included in geo scoring');
+const nodePrime = { public_key: 'ab8888', name: 'PrimeMeridian', lat: 10, lon: 0 };
+const anchorNearPM = { public_key: 'cd9999', name: 'AnchorPM', lat: 11, lon: 1 };
+HopResolver.init([nodePrime, nodeFar, anchorNearPM]);
+HopResolver.setAffinity({});
+const result8 = HopResolver.resolve(['cd99', 'ab'], 11.0, 1.0, null, null, null);
+assert(result8['ab'].name === 'PrimeMeridian',
+  'lon=0 candidate should be included and win by geo — got: ' + result8['ab'].name);
+
 console.log('\n' + (passed + failed) + ' tests, ' + passed + ' passed, ' + failed + ' failed\n');
 process.exit(failed > 0 ? 1 : 0);