diff --git a/CLAUDE.md b/CLAUDE.md
new file mode 100644
index 0000000..e7be5d2
--- /dev/null
+++ b/CLAUDE.md
@@ -0,0 +1,63 @@
+# CLAUDE.md
+
+This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.
+
+## Commands
+
+```bash
+# Build
+go build -o importer .
+
+# Build with version baked in
+go build -ldflags="-X main.version=v1.0.0" -o importer .
+
+# Run locally (requires IMPORT_DIR and LIBRARY_DIR env vars)
+IMPORT_DIR=/path/to/import LIBRARY_DIR=/path/to/library ./importer
+
+# Build Docker image
+docker build -t music-importer .
+
+# Build Docker image with version
+docker build --build-arg VERSION=v1.0.0 -t music-importer .
+```
+
+There are no tests in this codebase.
+
+## Architecture
+
+This is a single-package Go web app (`package main`) that runs as a web server on port 8080. Users trigger an import via the web UI, which runs the import pipeline in a background goroutine.
+
+**Pipeline flow** (`importer.go: RunImporter`):
+1. **Cluster** — loose audio files at the top of `IMPORT_DIR` are grouped into subdirectories by album tag (`files.go: cluster`)
+2. For each album directory:
+   - **Clean tags** — removes COMMENT/DESCRIPTION tags via `metaflac` (`audio.go`)
+   - **Tag metadata** — tries `beets` first; falls back to reading existing file tags, then MusicBrainz API (`metadata.go: getAlbumMetadata`)
+   - **Lyrics** — fetches synced LRC lyrics from LRClib API; falls back to plain lyrics formatted as LRC (`lrc.go`)
+   - **ReplayGain** — runs `rsgain easy` on the directory (`audio.go`)
+   - **Cover art** — looks for existing image files, downloads from Cover Art Archive via MusicBrainz if missing, then embeds into tracks (`media.go`)
+   - **Move** — moves tracks, .lrc files, and cover image into `LIBRARY_DIR/{Artist}/[{Date}] {Album} [{Quality}]/` (`files.go: moveToLibrary`)
+
+**Key types** (`importer.go`):
+- `AlbumResult` — tracks per-step success/failure/skip for one album
+- `ImportSession` — holds all `AlbumResult`s for one run; stored in `lastSession` global
+- `MusicMetadata` — artist/album/title/date/quality used throughout the pipeline
+
+**Web layer** (`main.go`):
+- `GET /` — renders `index.html.tmpl` with the last session's results
+- `POST /run` — starts `RunImporter()` in a goroutine; prevents concurrent runs via `importerMu` mutex
+
+**External tool dependencies** (must be present in PATH at runtime):
+- `ffprobe` — reads audio tags and stream info
+- `beet` — metadata tagging via MusicBrainz (primary metadata source)
+- `rsgain` — ReplayGain calculation
+- `metaflac` — FLAC tag manipulation and cover embedding
+- `curl` — MusicBrainz API fallback queries
+
+**Environment variables**:
+- `IMPORT_DIR` — source directory scanned for albums
+- `LIBRARY_DIR` — destination library root
+- `COPYMODE=true` — copies files instead of moving (still destructive on the destination)
+- `SLSKD_URL` — base URL of the slskd instance (e.g. `http://localhost:5030`)
+- `SLSKD_API_KEY` — slskd API key (sent as `X-API-Key` header)
+
+**Releases**: Docker image `gabehf/music-importer` is built and pushed to Docker Hub via GitHub Actions on `v*` tags.
diff --git a/README.md b/README.md
index c203fb1..52ac42a 100644
--- a/README.md
+++ b/README.md
@@ -1,5 +1,9 @@
 # music-importer
 
+Goes through a folder with a bunch of loose .flac/.mp3 files, or with album folders containing music files, then
+fetches metadata with beets/musicbrainz, downloads lyrics via LRClib, embeds discovered cover art, and moves them
+into the library with the format {Artist}/[{Year}] {Title} [{Format-Quality}]
+
 this thing is like 95% AI code. use at your own risk
 
 i didn't feel like spending the time to do it all right and i figured its simple enough that chatgpt couldn't possible screw it up *that* bad
@@ -21,6 +25,7 @@ services:
     environment:
       IMPORT_DIR: /import
       LIBRARY_DIR: /library
+      COPYMODE: true # copies files instead of moving. NOT NON-DESTRUCTIVE!!
 
 ```
 
diff --git a/discover.go b/discover.go
new file mode 100644
index 0000000..ddd38e6
--- /dev/null
+++ b/discover.go
@@ -0,0 +1,547 @@
+package main
+
+import (
+	"encoding/json"
+	"fmt"
+	"log"
+	"net/http"
+	"net/url"
+	"strings"
+	"sync"
+	"time"
+)
+
+// ── MusicBrainz types ─────────────────────────────────────────────────────────
+
+type mbArtistCredit struct {
+	Name   string `json:"name"`
+	Artist struct {
+		ID   string `json:"id"`
+		Name string `json:"name"`
+	} `json:"artist"`
+}
+
+type mbMedia struct {
+	Format     string `json:"format"`
+	TrackCount int    `json:"track-count"`
+}
+
+type mbRelease struct {
+	ID             string           `json:"id"`
+	Title          string           `json:"title"`
+	Date           string           `json:"date"`
+	Country        string           `json:"country"`
+	Disambiguation string           `json:"disambiguation"`
+	TextRepresentation struct {
+		Language string `json:"language"`
+	} `json:"text-representation"`
+	Media        []mbMedia        `json:"media"`
+	ArtistCredit []mbArtistCredit `json:"artist-credit"`
+	ReleaseGroup struct {
+		PrimaryType string `json:"primary-type"`
+	} `json:"release-group"`
+}
+
+type mbArtist struct {
+	ID             string `json:"id"`
+	Name           string `json:"name"`
+	Country        string `json:"country"`
+	Disambiguation string `json:"disambiguation"`
+}
+
+type mbReleaseGroup struct {
+	ID               string `json:"id"`
+	Title            string `json:"title"`
+	PrimaryType      string `json:"primary-type"`
+	FirstReleaseDate string `json:"first-release-date"`
+}
+
+// releaseTrackCount returns the total number of tracks across all media in a release.
+func releaseTrackCount(r mbRelease) int {
+	total := 0
+	for _, m := range r.Media {
+		total += m.TrackCount
+	}
+	return total
+}
+
+// getMBRelease fetches a single release by MBID (with media/track-count included).
+func getMBRelease(mbid string) (*mbRelease, error) {
+	var r mbRelease
+	err := mbGet(fmt.Sprintf("/ws/2/release/%s?fmt=json&inc=media", url.QueryEscape(mbid)), &r)
+	return &r, err
+}
+
+func mbGet(path string, out interface{}) error {
+	req, err := http.NewRequest("GET", "https://musicbrainz.org"+path, nil)
+	if err != nil {
+		return err
+	}
+	req.Header.Set("User-Agent", "music-importer/1.0 (https://github.com/gabehf/music-importer)")
+
+	resp, err := http.DefaultClient.Do(req)
+	if err != nil {
+		return err
+	}
+	defer resp.Body.Close()
+
+	if resp.StatusCode != http.StatusOK {
+		return fmt.Errorf("MusicBrainz returned %d", resp.StatusCode)
+	}
+	return json.NewDecoder(resp.Body).Decode(out)
+}
+
+func searchMBReleases(query string) ([]mbRelease, error) {
+	var result struct {
+		Releases []mbRelease `json:"releases"`
+	}
+	err := mbGet("/ws/2/release/?query="+url.QueryEscape(query)+"&fmt=json&limit=20&inc=media", &result)
+	return result.Releases, err
+}
+
+func searchMBArtists(query string) ([]mbArtist, error) {
+	var result struct {
+		Artists []mbArtist `json:"artists"`
+	}
+	err := mbGet("/ws/2/artist/?query="+url.QueryEscape(query)+"&fmt=json&limit=20", &result)
+	return result.Artists, err
+}
+
+// releaseFormatScore returns a preference score for a release's media format.
+// Higher is better. CD=2, Digital Media=1, anything else=0.
+func releaseFormatScore(r mbRelease) int {
+	for _, m := range r.Media {
+		switch m.Format {
+		case "Digital Media":
+			return 2
+		case "CD":
+			return 1
+		}
+	}
+	return 0
+}
+
+// releaseCountryScore returns a preference score for a release's country.
+// Higher is better. KR=3, JP=2, XW=1, anything else=0.
+func releaseCountryScore(r mbRelease) int {
+	switch r.Country {
+	case "XW":
+		return 2
+	case "KR":
+		return 1
+	}
+	return 0
+}
+
+// returns true if strings formatted 'YYYY-MM-DD" ts1 is before ts2
+func timeStringIsBefore(ts1, ts2 string) (bool, error) {
+	datefmt := "2006-02-01"
+	t1, err := time.Parse(datefmt, ts1)
+	if err != nil {
+		return false, err
+	}
+	t2, err := time.Parse(datefmt, ts2)
+	if err != nil {
+		return false, err
+	}
+	return t1.Unix() <= t2.Unix(), nil
+}
+
+// pickBestRelease selects the preferred release from a list.
+// No disambiguation (canonical release) is the primary sort key;
+// format (CD > Digital Media > *) is secondary; country (KR > XW > *) breaks ties.
+func pickBestRelease(releases []mbRelease) *mbRelease {
+	if len(releases) == 0 {
+		return nil
+	}
+	best := &releases[0]
+	for i := 1; i < len(releases); i++ {
+		r := &releases[i]
+
+		rNoDisamb := r.Disambiguation == ""
+		bestNoDisamb := best.Disambiguation == ""
+
+		// Prefer releases with no disambiguation — they are the canonical default.
+		if rNoDisamb && !bestNoDisamb {
+			best = r
+			continue
+		}
+		if !rNoDisamb && bestNoDisamb {
+			continue
+		}
+
+		// Both have the same disambiguation status; use date/format/country.
+		if before, err := timeStringIsBefore(r.Date, best.Date); before && err == nil {
+			rf, bf := releaseFormatScore(*r), releaseFormatScore(*best)
+			if rf > bf || (rf == bf && releaseCountryScore(*r) > releaseCountryScore(*best)) {
+				best = r
+			}
+		}
+	}
+	return best
+}
+
+// pickBestReleaseForGroup fetches all releases for a release group via the
+// MusicBrainz browse API (with media info) and returns the preferred release.
+// Returns nil on error or when the group has no releases.
+func pickBestReleaseForGroup(rgMBID string) *mbRelease {
+	var result struct {
+		Releases []mbRelease `json:"releases"`
+	}
+	path := fmt.Sprintf("/ws/2/release?release-group=%s&fmt=json&inc=media&limit=100", url.QueryEscape(rgMBID))
+	if err := mbGet(path, &result); err != nil || len(result.Releases) == 0 {
+		return nil
+	}
+	return pickBestRelease(result.Releases)
+}
+
+// getMBArtistReleaseGroups returns all Album and EP release groups for an artist,
+// paginating through the MusicBrainz browse API with the required 1 req/s rate limit.
+func getMBArtistReleaseGroups(artistMBID string) ([]mbReleaseGroup, error) {
+	const limit = 100
+	var all []mbReleaseGroup
+
+	for offset := 0; ; offset += limit {
+		path := fmt.Sprintf(
+			"/ws/2/release-group?artist=%s&type=album%%7Cep&fmt=json&limit=%d&offset=%d",
+			url.QueryEscape(artistMBID), limit, offset,
+		)
+
+		var result struct {
+			ReleaseGroups []mbReleaseGroup `json:"release-groups"`
+			Count         int              `json:"release-group-count"`
+		}
+		if err := mbGet(path, &result); err != nil {
+			return all, err
+		}
+
+		for _, rg := range result.ReleaseGroups {
+			t := strings.ToLower(rg.PrimaryType)
+			if t == "album" || t == "ep" {
+				all = append(all, rg)
+			}
+		}
+
+		if offset+limit >= result.Count {
+			break
+		}
+		time.Sleep(time.Second) // MusicBrainz rate limit
+	}
+
+	return all, nil
+}
+
+// fetchArtist fetches every Album and EP release group for an artist by running
+// fetchRelease for each one sequentially, then registers each for monitoring.
+func fetchArtist(artistMBID, artistName string, logf func(string)) error {
+	log.Printf("[discover] artist fetch started: %s (%s)", artistName, artistMBID)
+	logf(fmt.Sprintf("Looking up discography for %s on MusicBrainz…", artistName))
+
+	groups, err := getMBArtistReleaseGroups(artistMBID)
+	if err != nil {
+		return fmt.Errorf("MusicBrainz discography lookup failed: %w", err)
+	}
+	if len(groups) == 0 {
+		return fmt.Errorf("no albums or EPs found for %s on MusicBrainz", artistName)
+	}
+
+	log.Printf("[discover] found %d release groups for %s", len(groups), artistName)
+	logf(fmt.Sprintf("Found %d albums/EPs", len(groups)))
+
+	failed := 0
+	for i, rg := range groups {
+		logf(fmt.Sprintf("[%d/%d] %s", i+1, len(groups), rg.Title))
+		// Pick the best release for this group. beets --search-id requires a
+		// release MBID; release group MBIDs are not accepted.
+		time.Sleep(time.Second) // MusicBrainz rate limit
+		rel := pickBestReleaseForGroup(rg.ID)
+		releaseMBID := ""
+		trackCount := 0
+		if rel == nil {
+			logf(fmt.Sprintf("  ↳ warning: could not resolve release for group %s, beets will search by name", rg.ID))
+		} else {
+			releaseMBID = rel.ID
+			trackCount = releaseTrackCount(*rel)
+			format := ""
+			if len(rel.Media) > 0 {
+				format = rel.Media[0].Format
+			}
+			logf(fmt.Sprintf("  ↳ selected release: %s [%s / %s / %d tracks]", releaseMBID, format, rel.Country, trackCount))
+		}
+
+		folder, err := fetchRelease(artistName, rg.Title, releaseMBID, trackCount, logf)
+		if err != nil {
+			log.Printf("[discover] fetch failed for %q by %s: %v", rg.Title, artistName, err)
+			logf(fmt.Sprintf("  ↳ failed: %v", err))
+			failed++
+			continue
+		}
+		// Key the pending download by release group ID for dedup; beets uses releaseMBID.
+		registerDownload(rg.ID, releaseMBID, artistName, rg.Title, trackCount, folder, nil)
+		logf(fmt.Sprintf("  ↳ registered for import (release mbid: %s)", releaseMBID))
+	}
+
+	if failed > 0 {
+		logf(fmt.Sprintf("Done — %d/%d queued, %d failed", len(groups)-failed, len(groups), failed))
+	} else {
+		logf(fmt.Sprintf("Done — all %d downloads queued, monitoring for import", len(groups)))
+	}
+	log.Printf("[discover] artist fetch complete: %s (%d/%d succeeded)", artistName, len(groups)-failed, len(groups))
+	return nil
+}
+
+// ── Fetch state ───────────────────────────────────────────────────────────────
+
+type fetchEntry struct {
+	mu      sync.Mutex
+	ID      string   `json:"id"`
+	Artist  string   `json:"artist"`
+	Album   string   `json:"album"`
+	Log     []string `json:"log"`
+	Done    bool     `json:"done"`
+	Success bool     `json:"success"`
+	ErrMsg  string   `json:"error,omitempty"`
+}
+
+var (
+	fetchesMu sync.Mutex
+	fetchMap  = make(map[string]*fetchEntry)
+)
+
+func newFetchEntry(id, artist, album string) *fetchEntry {
+	e := &fetchEntry{ID: id, Artist: artist, Album: album}
+	fetchesMu.Lock()
+	fetchMap[id] = e
+	fetchesMu.Unlock()
+	return e
+}
+
+func (e *fetchEntry) appendLog(msg string) {
+	e.mu.Lock()
+	e.Log = append(e.Log, msg)
+	e.mu.Unlock()
+}
+
+func (e *fetchEntry) finish(err error) {
+	e.mu.Lock()
+	e.Done = true
+	if err != nil {
+		e.ErrMsg = err.Error()
+	} else {
+		e.Success = true
+	}
+	e.mu.Unlock()
+}
+
+func (e *fetchEntry) snapshot() fetchEntry {
+	e.mu.Lock()
+	defer e.mu.Unlock()
+	cp := *e
+	cp.Log = append([]string(nil), e.Log...)
+	return cp
+}
+
+// ── HTTP handlers ─────────────────────────────────────────────────────────────
+
+// handleDiscoverSearch handles GET /discover/search?q=...&type=release|artist
+func handleDiscoverSearch(w http.ResponseWriter, r *http.Request) {
+	q := r.URL.Query().Get("q")
+	if q == "" {
+		http.Error(w, "missing q", http.StatusBadRequest)
+		return
+	}
+	searchType := r.URL.Query().Get("type")
+	if searchType == "" {
+		searchType = "release"
+	}
+	log.Printf("[discover] search: type=%s q=%q", searchType, q)
+
+	w.Header().Set("Content-Type", "application/json")
+
+	switch searchType {
+	case "artist":
+		artists, err := searchMBArtists(q)
+		if err != nil {
+			log.Printf("[discover] artist search error: %v", err)
+			http.Error(w, err.Error(), http.StatusInternalServerError)
+			return
+		}
+		log.Printf("[discover] artist search returned %d results", len(artists))
+		json.NewEncoder(w).Encode(artists)
+
+	default: // "release"
+		releases, err := searchMBReleases(q)
+		if err != nil {
+			log.Printf("[discover] release search error: %v", err)
+			http.Error(w, err.Error(), http.StatusInternalServerError)
+			return
+		}
+		log.Printf("[discover] release search returned %d results", len(releases))
+		json.NewEncoder(w).Encode(releases)
+	}
+}
+
+// handleDiscoverFetch handles POST /discover/fetch
+// Body: {"id":"mbid","artist":"...","album":"..."}
+func handleDiscoverFetch(w http.ResponseWriter, r *http.Request) {
+	if r.Method != http.MethodPost {
+		http.Error(w, "POST only", http.StatusMethodNotAllowed)
+		return
+	}
+
+	var body struct {
+		ID     string `json:"id"`
+		Artist string `json:"artist"`
+		Album  string `json:"album"`
+	}
+	if err := json.NewDecoder(r.Body).Decode(&body); err != nil || body.ID == "" || body.Artist == "" || body.Album == "" {
+		http.Error(w, "id, artist and album are required", http.StatusBadRequest)
+		return
+	}
+
+	// If a fetch for this ID is already in progress, return its ID without starting a new one.
+	fetchesMu.Lock()
+	existing := fetchMap[body.ID]
+	fetchesMu.Unlock()
+	if existing != nil {
+		existing.mu.Lock()
+		done := existing.Done
+		existing.mu.Unlock()
+		if !done {
+			w.Header().Set("Content-Type", "application/json")
+			json.NewEncoder(w).Encode(map[string]string{"id": body.ID})
+			return
+		}
+	}
+
+	log.Printf("[discover] starting fetch: %q by %s (mbid: %s)", body.Album, body.Artist, body.ID)
+	entry := newFetchEntry(body.ID, body.Artist, body.Album)
+
+	// Look up the expected track count from MusicBrainz so the folder-selection
+	// logic can prefer results that match the release we intend to import.
+	trackCount := 0
+	if rel, err := getMBRelease(body.ID); err == nil {
+		trackCount = releaseTrackCount(*rel)
+		log.Printf("[discover] release %s has %d tracks", body.ID, trackCount)
+	} else {
+		log.Printf("[discover] could not fetch release track count for %s: %v", body.ID, err)
+	}
+
+	go func() {
+		folder, err := fetchRelease(body.Artist, body.Album, body.ID, trackCount, entry.appendLog)
+		if err != nil {
+			log.Printf("[discover] fetch failed for %q by %s: %v", body.Album, body.Artist, err)
+			entry.finish(err)
+			return
+		}
+		log.Printf("[discover] fetch complete for %q by %s, registering for import", body.Album, body.Artist)
+		registerDownload(body.ID, body.ID, body.Artist, body.Album, trackCount, folder, entry)
+		// entry.finish is called by the monitor when import completes
+	}()
+
+	w.Header().Set("Content-Type", "application/json")
+	json.NewEncoder(w).Encode(map[string]string{"id": body.ID})
+}
+
+// handleDiscoverFetchArtist handles POST /discover/fetch/artist
+// Body: {"id":"artist-mbid","name":"Artist Name"}
+func handleDiscoverFetchArtist(w http.ResponseWriter, r *http.Request) {
+	if r.Method != http.MethodPost {
+		http.Error(w, "POST only", http.StatusMethodNotAllowed)
+		return
+	}
+
+	var body struct {
+		ID   string `json:"id"`
+		Name string `json:"name"`
+	}
+	if err := json.NewDecoder(r.Body).Decode(&body); err != nil || body.ID == "" || body.Name == "" {
+		http.Error(w, "id and name are required", http.StatusBadRequest)
+		return
+	}
+
+	fetchesMu.Lock()
+	existing := fetchMap[body.ID]
+	fetchesMu.Unlock()
+	if existing != nil {
+		existing.mu.Lock()
+		done := existing.Done
+		existing.mu.Unlock()
+		if !done {
+			w.Header().Set("Content-Type", "application/json")
+			json.NewEncoder(w).Encode(map[string]string{"id": body.ID})
+			return
+		}
+	}
+
+	log.Printf("[discover] starting artist fetch: %s (%s)", body.Name, body.ID)
+	entry := newFetchEntry(body.ID, body.Name, "")
+	go func() {
+		err := fetchArtist(body.ID, body.Name, entry.appendLog)
+		if err != nil {
+			log.Printf("[discover] artist fetch failed for %s: %v", body.Name, err)
+		} else {
+			log.Printf("[discover] artist fetch complete for %s", body.Name)
+		}
+		entry.finish(err)
+	}()
+
+	w.Header().Set("Content-Type", "application/json")
+	json.NewEncoder(w).Encode(map[string]string{"id": body.ID})
+}
+
+// handleDiscoverFetchStatus handles GET /discover/fetch/status?id=...
+func handleDiscoverFetchStatus(w http.ResponseWriter, r *http.Request) {
+	id := r.URL.Query().Get("id")
+	if id == "" {
+		http.Error(w, "missing id", http.StatusBadRequest)
+		return
+	}
+
+	fetchesMu.Lock()
+	entry := fetchMap[id]
+	fetchesMu.Unlock()
+
+	if entry == nil {
+		http.Error(w, "not found", http.StatusNotFound)
+		return
+	}
+
+	snap := entry.snapshot()
+	w.Header().Set("Content-Type", "application/json")
+	json.NewEncoder(w).Encode(snap)
+}
+
+// fetchListItem is a summary of a fetch entry for the list endpoint.
+type fetchListItem struct {
+	ID      string `json:"id"`
+	Title   string `json:"title"`
+	Done    bool   `json:"done"`
+	Success bool   `json:"success"`
+}
+
+// handleDiscoverFetchList handles GET /discover/fetch/list
+// Returns a summary of all known fetch entries so the frontend can discover
+// entries created server-side (e.g. per-album entries from an artist fetch).
+func handleDiscoverFetchList(w http.ResponseWriter, r *http.Request) {
+	fetchesMu.Lock()
+	items := make([]fetchListItem, 0, len(fetchMap))
+	for _, e := range fetchMap {
+		e.mu.Lock()
+		title := e.Artist
+		if e.Album != "" {
+			title = e.Artist + " \u2014 " + e.Album
+		}
+		items = append(items, fetchListItem{
+			ID:      e.ID,
+			Title:   title,
+			Done:    e.Done,
+			Success: e.Success,
+		})
+		e.mu.Unlock()
+	}
+	fetchesMu.Unlock()
+
+	w.Header().Set("Content-Type", "application/json")
+	json.NewEncoder(w).Encode(items)
+}
diff --git a/files.go b/files.go
index 5155997..75a388c 100644
--- a/files.go
+++ b/files.go
@@ -2,22 +2,41 @@ package main
 
 import (
 	"fmt"
+	"io"
 	"os"
 	"path"
 	"path/filepath"
 	"strings"
 )
 
-// moveToLibrary moves a file to {libDir}/{artist}/[{year}] {album}/filename.
+// albumTargetDir returns the destination directory for an album without
+// creating it. Use this to check for an existing import before moving files.
+func albumTargetDir(libDir string, md *MusicMetadata) string {
+	date := md.Date
+	if date == "" {
+		date = md.Year
+	}
+	albumDir := fmt.Sprintf("[%s] %s", date, md.Album)
+	if md.Quality != "" {
+		albumDir += fmt.Sprintf(" [%s]", md.Quality)
+	}
+	return filepath.Join(libDir, sanitize(md.Artist), sanitize(albumDir))
+}
+
+// moveToLibrary moves a file to {libDir}/{artist}/[{date}] {album} [{quality}]/filename.
 func moveToLibrary(libDir string, md *MusicMetadata, srcPath string) error {
-	targetDir := filepath.Join(libDir, sanitize(md.Artist), sanitize(fmt.Sprintf("[%s] %s", md.Year, md.Album)))
+	targetDir := albumTargetDir(libDir, md)
 	if err := os.MkdirAll(targetDir, 0755); err != nil {
 		return err
 	}
 
 	dst := filepath.Join(targetDir, filepath.Base(srcPath))
 	fmt.Println("→ Moving:", srcPath, "→", dst)
-	return os.Rename(srcPath, dst)
+	if strings.ToLower(os.Getenv("COPYMODE")) == "true" {
+		return copy(srcPath, dst)
+	} else {
+		return os.Rename(srcPath, dst)
+	}
 }
 
 // cluster moves all top-level audio files in dir into subdirectories named
@@ -101,3 +120,63 @@ func sanitize(s string) string {
 	)
 	return r.Replace(s)
 }
+
+// CopyFile copies a file from src to dst. If src and dst files exist, and are
+// the same, then return success. Otherise, attempt to create a hard link
+// between the two files. If that fail, copy the file contents from src to dst.
+func copy(src, dst string) (err error) {
+	sfi, err := os.Stat(src)
+	if err != nil {
+		return
+	}
+	if !sfi.Mode().IsRegular() {
+		// cannot copy non-regular files (e.g., directories,
+		// symlinks, devices, etc.)
+		return fmt.Errorf("CopyFile: non-regular source file %s (%q)", sfi.Name(), sfi.Mode().String())
+	}
+	dfi, err := os.Stat(dst)
+	if err != nil {
+		if !os.IsNotExist(err) {
+			return
+		}
+	} else {
+		if !(dfi.Mode().IsRegular()) {
+			return fmt.Errorf("CopyFile: non-regular destination file %s (%q)", dfi.Name(), dfi.Mode().String())
+		}
+		if os.SameFile(sfi, dfi) {
+			return
+		}
+	}
+	if err = os.Link(src, dst); err == nil {
+		return
+	}
+	err = copyFileContents(src, dst)
+	return
+}
+
+// copyFileContents copies the contents of the file named src to the file named
+// by dst. The file will be created if it does not already exist. If the
+// destination file exists, all it's contents will be replaced by the contents
+// of the source file.
+func copyFileContents(src, dst string) (err error) {
+	in, err := os.Open(src)
+	if err != nil {
+		return
+	}
+	defer in.Close()
+	out, err := os.Create(dst)
+	if err != nil {
+		return
+	}
+	defer func() {
+		cerr := out.Close()
+		if err == nil {
+			err = cerr
+		}
+	}()
+	if _, err = io.Copy(out, in); err != nil {
+		return
+	}
+	err = out.Sync()
+	return
+}
diff --git a/importer.go b/importer.go
index ac449b5..d108f16 100644
--- a/importer.go
+++ b/importer.go
@@ -5,8 +5,116 @@ import (
 	"log"
 	"os"
 	"path/filepath"
+	"time"
 )
 
+// StepStatus records the outcome of a single pipeline step for an album.
+type StepStatus struct {
+	Skipped bool
+	Err     error
+}
+
+func (s StepStatus) Failed() bool { return s.Err != nil }
+
+// MetadataSource identifies which backend resolved the album metadata.
+type MetadataSource string
+
+const (
+	MetadataSourceBeets       MetadataSource = "beets"
+	MetadataSourceMusicBrainz MetadataSource = "musicbrainz"
+	MetadataSourceFileTags    MetadataSource = "file_tags"
+	MetadataSourceUnknown     MetadataSource = ""
+)
+
+// LyricsStats summarises per-track lyric discovery for an album.
+type LyricsStats struct {
+	Total      int // total audio tracks examined
+	Synced     int // tracks with synced (timestamped) LRC lyrics downloaded
+	Plain      int // tracks with plain (un-timestamped) lyrics downloaded
+	AlreadyHad int // tracks that already had an .lrc file, skipped
+	NotFound   int // tracks for which no lyrics could be found
+}
+
+func (l LyricsStats) Downloaded() int { return l.Synced + l.Plain }
+
+// CoverArtStats records what happened with cover art for an album.
+type CoverArtStats struct {
+	Found    bool   // a cover image file was found in the folder
+	Embedded bool   // cover was successfully embedded into tracks
+	Source   string // filename of the cover image, e.g. "cover.jpg"
+}
+
+// AlbumResult holds the outcome of every pipeline step for one imported album.
+type AlbumResult struct {
+	Name     string
+	Path     string
+	Metadata *MusicMetadata
+
+	MetadataSource MetadataSource
+	LyricsStats    LyricsStats
+	CoverArtStats  CoverArtStats
+	TrackCount     int
+
+	CleanTags   StepStatus
+	TagMetadata StepStatus
+	Lyrics      StepStatus
+	ReplayGain  StepStatus
+	CoverArt    StepStatus
+	Move        StepStatus
+
+	// FatalStep is the name of the step that caused the album to be skipped
+	// entirely, or empty if the album completed the full pipeline.
+	FatalStep string
+}
+
+func (a *AlbumResult) skippedAt(step string) {
+	a.FatalStep = step
+}
+
+func (a *AlbumResult) Succeeded() bool { return a.FatalStep == "" }
+func (a *AlbumResult) HasWarnings() bool {
+	if a.CleanTags.Failed() ||
+		a.TagMetadata.Failed() ||
+		a.Lyrics.Failed() ||
+		a.ReplayGain.Failed() ||
+		a.CoverArt.Failed() ||
+		a.Move.Failed() {
+		return true
+	} else {
+		return false
+	}
+}
+
+// ImportSession holds the results of a single importer run.
+type ImportSession struct {
+	StartedAt  time.Time
+	FinishedAt time.Time
+	Albums     []*AlbumResult
+}
+
+func (s *ImportSession) Failed() []*AlbumResult {
+	var out []*AlbumResult
+	for _, a := range s.Albums {
+		if !a.Succeeded() {
+			out = append(out, a)
+		}
+	}
+	return out
+}
+
+func (s *ImportSession) WithWarnings() []*AlbumResult {
+	var out []*AlbumResult
+	for _, a := range s.Albums {
+		if a.Succeeded() && a.HasWarnings() {
+			out = append(out, a)
+		}
+	}
+	return out
+}
+
+// lastSession is populated at the end of each RunImporter call.
+var lastSession *ImportSession
+
 func RunImporter() {
 	importDir := os.Getenv("IMPORT_DIR")
 	libraryDir := os.Getenv("LIBRARY_DIR")
@@ -29,6 +137,12 @@ func RunImporter() {
 		return
 	}
 
+	session := &ImportSession{StartedAt: time.Now()}
+	defer func() {
+		session.FinishedAt = time.Now()
+		lastSession = session
+	}()
+
 	fmt.Println("=== Starting Import ===")
 
 	if err := cluster(importDir); err != nil {
@@ -60,59 +174,102 @@ func RunImporter() {
 
 		fmt.Println("\n===== Album:", e.Name(), "=====")
 
+		result := &AlbumResult{Name: e.Name(), Path: albumPath}
+		session.Albums = append(session.Albums, result)
+		result.TrackCount = len(tracks)
+
 		fmt.Println("→ Cleaning album tags:")
-		if err = cleanAlbumTags(albumPath); err != nil {
-			fmt.Println("Cleaning album tags failed:", err)
+		result.CleanTags.Err = cleanAlbumTags(albumPath)
+		if result.CleanTags.Failed() {
+			fmt.Println("Cleaning album tags failed:", result.CleanTags.Err)
 		}
 
 		fmt.Println("→ Tagging album metadata:")
-		md, err := getAlbumMetadata(albumPath, tracks[0])
+		md, src, err := getAlbumMetadata(albumPath, tracks[0], "")
+		result.TagMetadata.Err = err
+		result.MetadataSource = src
 		if err != nil {
 			fmt.Println("Metadata failed, skipping album:", err)
+			result.skippedAt("TagMetadata")
 			continue
 		}
+		result.Metadata = md
 
 		fmt.Println("→ Fetching synced lyrics from LRCLIB:")
-		if err := DownloadAlbumLyrics(albumPath); err != nil {
+		lyricsStats, err := DownloadAlbumLyrics(albumPath)
+		result.Lyrics.Err = err
+		result.LyricsStats = lyricsStats
+		if result.Lyrics.Failed() {
 			fmt.Println("Failed to download synced lyrics.")
 		}
 
 		fmt.Println("→ Applying ReplayGain to album:", albumPath)
-		if err := applyReplayGain(albumPath); err != nil {
-			fmt.Println("ReplayGain failed, skipping album:", err)
+		result.ReplayGain.Err = applyReplayGain(albumPath)
+		if result.ReplayGain.Failed() {
+			fmt.Println("ReplayGain failed, skipping album:", result.ReplayGain.Err)
+			result.skippedAt("ReplayGain")
 			continue
 		}
 
+		fmt.Println("→ Downloading cover art for album:", albumPath)
+		if _, err := FindCoverImage(albumPath); err != nil {
+			if err := DownloadCoverArt(albumPath, md, ""); err != nil {
+				fmt.Println("Cover art download failed:", err)
+			}
+		}
+
+		if err := NormalizeCoverArt(albumPath); err != nil {
+			fmt.Println("Cover art normalization warning:", err)
+		}
+
 		fmt.Println("→ Embedding cover art for album:", albumPath)
-		if err := EmbedAlbumArtIntoFolder(albumPath); err != nil {
-			fmt.Println("Cover embed failed, skipping album:", err)
+		result.CoverArt.Err = EmbedAlbumArtIntoFolder(albumPath)
+		if coverImg, err := FindCoverImage(albumPath); err == nil {
+			result.CoverArtStats.Found = true
+			result.CoverArtStats.Source = filepath.Base(coverImg)
+			if result.CoverArt.Err == nil {
+				result.CoverArtStats.Embedded = true
+			}
+		}
+		if result.CoverArt.Failed() {
+			fmt.Println("Cover embed failed, skipping album:", result.CoverArt.Err)
+			result.skippedAt("CoverArt")
 			continue
 		}
 
-		fmt.Println("→ Moving tracks into library for album:", albumPath)
-		for _, track := range tracks {
-			if err := moveToLibrary(libraryDir, md, track); err != nil {
-				fmt.Println("Failed to move track:", track, err)
+		targetDir := albumTargetDir(libraryDir, md)
+		if _, err := os.Stat(targetDir); err == nil {
+			fmt.Println("→ Album already exists in library, skipping move:", targetDir)
+			result.Move.Skipped = true
+		} else {
+			fmt.Println("→ Moving tracks into library for album:", albumPath)
+			for _, track := range tracks {
+				if err := moveToLibrary(libraryDir, md, track); err != nil {
+					fmt.Println("Failed to move track:", track, err)
+					result.Move.Err = err // retains last error; all attempts are still made
+				}
 			}
-		}
 
-		lyrics, _ := getLyricFiles(albumPath)
+			lyrics, _ := getLyricFiles(albumPath)
 
-		fmt.Println("→ Moving lyrics into library for album:", albumPath)
-		for _, file := range lyrics {
-			if err := moveToLibrary(libraryDir, md, file); err != nil {
-				fmt.Println("Failed to move lyrics:", file, err)
+			fmt.Println("→ Moving lyrics into library for album:", albumPath)
+			for _, file := range lyrics {
+				if err := moveToLibrary(libraryDir, md, file); err != nil {
+					fmt.Println("Failed to move lyrics:", file, err)
+					result.Move.Err = err
+				}
 			}
-		}
 
-		fmt.Println("→ Moving album cover into library for album:", albumPath)
-		if coverImg, err := FindCoverImage(albumPath); err == nil {
-			if err := moveToLibrary(libraryDir, md, coverImg); err != nil {
-				fmt.Println("Failed to cover image:", coverImg, err)
+			fmt.Println("→ Moving album cover into library for album:", albumPath)
+			if coverImg, err := FindCoverImage(albumPath); err == nil {
+				if err := moveToLibrary(libraryDir, md, coverImg); err != nil {
+					fmt.Println("Failed to cover image:", coverImg, err)
+					result.Move.Err = err
+				}
 			}
-		}
 
-		os.Remove(albumPath)
+			os.Remove(albumPath)
+		}
 	}
 
 	fmt.Println("\n=== Import Complete ===")
diff --git a/index.html.tmpl b/index.html.tmpl
new file mode 100644
index 0000000..f89241c
--- /dev/null
+++ b/index.html.tmpl
@@ -0,0 +1,143 @@
+<!DOCTYPE html>
+<html lang="en">
+<head>
+	<meta charset="UTF-8">
+	<meta name="viewport" content="width=device-width, initial-scale=1">
+	<title>Music Importer</title>
+	<link rel="stylesheet" href="/static/style.css?v={{.Version}}">
+</head>
+<body>
+	<h1>Music Importer</h1>
+
+	<nav class="tabs">
+		<button class="tab-btn active" data-tab="import">Import</button>
+		<button class="tab-btn" data-tab="discover">Discover</button>
+	</nav>
+
+	<!-- ── Import ─────────────────────────────────────────────────────────── -->
+	<section id="tab-import" class="tab-pane active">
+		<form action="/run" method="POST">
+			<button type="submit" class="run-btn" {{if .Running}}disabled{{end}}>
+				{{if .Running}}Importer Running…{{else}}Run Importer{{end}}
+			</button>
+		</form>
+
+		{{with .Session}}
+		<div class="content-box session">
+			<div class="session-header">
+				<h2>Last Run &mdash; {{.StartedAt.Format "Jan 2, 2006 15:04:05"}}</h2>
+				<span class="duration">{{duration .StartedAt .FinishedAt}}</span>
+			</div>
+
+			{{range .Albums}}{{$album := .}}
+			<article class="album">
+				<div class="album-header">
+					<span class="album-name" title="{{.Path}}">{{.Name}}</span>
+					{{if .Succeeded}}
+						{{if .HasWarnings}}
+							<span class="badge badge-warn">&#9888; warnings</span>
+						{{else}}
+							<span class="badge badge-ok">&#10003; ok</span>
+						{{end}}
+					{{else}}
+						<span class="badge badge-fatal">&#10007; failed at {{.FatalStep}}</span>
+					{{end}}
+				</div>
+
+				{{with .Metadata}}
+				<div class="metadata">
+					<span class="metadata-title">{{.Artist}} &mdash; {{.Album}}{{if .Year}} ({{.Year}}){{end}}</span>
+					{{if $album.MetadataSource}}
+					<span class="metadata-pill">
+						<span class="pill-label">via</span>
+						{{if eq (print $album.MetadataSource) "beets"}}
+							<span class="pill-beets">beets</span>
+						{{else if eq (print $album.MetadataSource) "musicbrainz"}}
+							<span class="pill-musicbrainz">MusicBrainz</span>
+						{{else if eq (print $album.MetadataSource) "file_tags"}}
+							<span class="pill-file_tags">file tags</span>
+						{{else}}
+							<span class="pill-unknown">unknown</span>
+						{{end}}
+					</span>
+					{{end}}
+				</div>
+				{{end}}
+
+				<div class="info-grid">
+					<div class="info-card">
+						<div class="info-card-label">Tracks</div>
+						<div class="info-card-value">{{.TrackCount}}</div>
+					</div>
+
+					<div class="info-card">
+						<div class="info-card-label">Lyrics</div>
+						{{if eq .LyricsStats.Total 0}}
+							<div class="info-card-value info-dim">n/a</div>
+						{{else}}
+							<div class="info-card-value {{if gt .LyricsStats.Downloaded 0}}info-ok{{else}}info-dim{{end}}">
+								{{.LyricsStats.Downloaded}} / {{.LyricsStats.Total}}
+							</div>
+							<div class="info-card-sub">
+								{{if gt .LyricsStats.Synced 0}}<span class="info-ok">{{.LyricsStats.Synced}} synced</span>{{end}}
+								{{if and (gt .LyricsStats.Synced 0) (gt .LyricsStats.Plain 0)}} &middot; {{end}}
+								{{if gt .LyricsStats.Plain 0}}<span class="info-warn">{{.LyricsStats.Plain}} plain</span>{{end}}
+								{{if gt .LyricsStats.AlreadyHad 0}}<span class="info-dim"> {{.LyricsStats.AlreadyHad}} existing</span>{{end}}
+								{{if gt .LyricsStats.NotFound 0}}<span class="info-dim"> {{.LyricsStats.NotFound}} missing</span>{{end}}
+							</div>
+						{{end}}
+					</div>
+
+					<div class="info-card">
+						<div class="info-card-label">Cover Art</div>
+						{{if .CoverArtStats.Found}}
+							{{if .CoverArtStats.Embedded}}
+								<div class="info-card-value info-ok">Embedded</div>
+								<div class="info-card-sub info-dim">{{.CoverArtStats.Source}}</div>
+							{{else}}
+								<div class="info-card-value info-warn">Found, not embedded</div>
+								<div class="info-card-sub info-dim">{{.CoverArtStats.Source}}</div>
+							{{end}}
+						{{else}}
+							<div class="info-card-value info-dim">Not found</div>
+						{{end}}
+					</div>
+				</div>
+
+				<div class="steps-label">Pipeline</div>
+				<div class="steps">
+					{{stepCell "Clean Tags" .CleanTags  ""}}
+					{{stepCell "Metadata"   .TagMetadata .FatalStep}}
+					{{stepCell "Lyrics"     .Lyrics      ""}}
+					{{stepCell "ReplayGain" .ReplayGain  .FatalStep}}
+					{{stepCell "Cover Art"  .CoverArt    .FatalStep}}
+					{{stepCell "Move"       .Move        ""}}
+				</div>
+			</article>
+			{{end}}
+		</div>
+		{{end}}
+	</section>
+
+	<!-- ── Discover ───────────────────────────────────────────────────────── -->
+	<section id="tab-discover" class="tab-pane">
+		<div class="content-box">
+			<div class="search-form">
+				<div class="type-toggle">
+					<button class="type-btn active" data-type="release">Release</button>
+					<button class="type-btn" data-type="artist">Artist</button>
+				</div>
+				<input id="search-q" class="search-input" type="search"
+					placeholder="Search MusicBrainz…" autocomplete="off">
+				<button id="search-btn" class="search-btn">Search</button>
+			</div>
+			<div id="search-results"></div>
+		</div>
+		<div class="content-box fetch-list" id="fetch-list"></div>
+	</section>
+
+	<footer>{{.Version}}</footer>
+
+	<script src="/static/app.js?v={{.Version}}" defer></script>
+</body>
+</html>
diff --git a/lrc.go b/lrc.go
index 94b1233..bdfae86 100644
--- a/lrc.go
+++ b/lrc.go
@@ -51,7 +51,8 @@ func TrackDuration(path string) (int, error) {
 
 // DownloadAlbumLyrics downloads synced lyrics (LRC format) for each track in the album directory.
 // Assumes metadata is already final (tags complete).
-func DownloadAlbumLyrics(albumDir string) error {
+func DownloadAlbumLyrics(albumDir string) (LyricsStats, error) {
+	var stats LyricsStats
 	err := filepath.Walk(albumDir, func(path string, info os.FileInfo, err error) error {
 		if err != nil {
 			return err
@@ -64,10 +65,12 @@ func DownloadAlbumLyrics(albumDir string) error {
 		if ext != ".mp3" && ext != ".flac" {
 			return nil
 		}
+		stats.Total++
 
 		// Skip if LRC already exists next to the file
 		lrcPath := strings.TrimSuffix(path, ext) + ".lrc"
 		if _, err := os.Stat(lrcPath); err == nil {
+			stats.AlreadyHad++
 			fmt.Println("→ Skipping (already has lyrics):", filepath.Base(path))
 			return nil
 		}
@@ -75,18 +78,21 @@ func DownloadAlbumLyrics(albumDir string) error {
 		// Read metadata
 		md, err := readTags(path)
 		if err != nil {
+			stats.NotFound++
 			fmt.Println("Skipping (unable to read tags):", path, "error:", err)
 			return nil
 		}
 		if md.Title == "" || md.Artist == "" || md.Album == "" {
+			stats.NotFound++
 			fmt.Println("Skipping (missing metadata):", path)
 			return nil
 		}
 
 		duration, _ := TrackDuration(path)
 
-		lyrics, err := fetchLRCLibLyrics(md.Artist, md.Title, md.Album, duration)
+		lyrics, synced, err := fetchLRCLibLyrics(md.Artist, md.Title, md.Album, duration)
 		if err != nil {
+			stats.NotFound++
 			fmt.Println("No lyrics found:", md.Artist, "-", md.Title)
 			return nil
 		}
@@ -96,15 +102,20 @@ func DownloadAlbumLyrics(albumDir string) error {
 			return fmt.Errorf("writing lrc file for %s: %w", path, err)
 		}
 
+		if synced {
+			stats.Synced++
+		} else {
+			stats.Plain++
+		}
 		fmt.Println("→ Downloaded lyrics:", filepath.Base(lrcPath))
 		return nil
 	})
 
-	return err
+	return stats, err
 }
 
 // fetchLRCLibLyrics calls the LRCLIB API and returns synced lyrics if available.
-func fetchLRCLibLyrics(artist, title, album string, duration int) (string, error) {
+func fetchLRCLibLyrics(artist, title, album string, duration int) (string, bool, error) {
 
 	url := fmt.Sprintf(
 		"https://lrclib.net/api/get?artist_name=%s&track_name=%s&album_name=%s&duration=%d",
@@ -113,35 +124,35 @@ func fetchLRCLibLyrics(artist, title, album string, duration int) (string, error
 
 	resp, err := http.Get(url)
 	if err != nil {
-		return "", fmt.Errorf("lrclib fetch error: %w", err)
+		return "", false, fmt.Errorf("lrclib fetch error: %w", err)
 	}
 	defer resp.Body.Close()
 
 	if resp.StatusCode != http.StatusOK {
-		return "", fmt.Errorf("lrclib returned status %d", resp.StatusCode)
+		return "", false, fmt.Errorf("lrclib returned status %d", resp.StatusCode)
 	}
 
 	bodyBytes, err := io.ReadAll(resp.Body)
 	if err != nil {
-		return "", fmt.Errorf("reading lrclib response: %w", err)
+		return "", false, fmt.Errorf("reading lrclib response: %w", err)
 	}
 
 	var out LRCLibResponse
 	if err := json.Unmarshal(bodyBytes, &out); err != nil {
-		return "", fmt.Errorf("parsing lrclib json: %w", err)
+		return "", false, fmt.Errorf("parsing lrclib json: %w", err)
 	}
 
 	if out.SyncedLyrics != "" {
-		return out.SyncedLyrics, nil
+		return out.SyncedLyrics, true, nil
 	}
 
 	// If no syncedLyrics, fallback to plain
 	if out.PlainLyrics != "" {
 		// Convert plain text to a fake LRC wrapper
-		return plainToLRC(out.PlainLyrics), nil
+		return plainToLRC(out.PlainLyrics), false, nil
 	}
 
-	return "", fmt.Errorf("no lyrics found")
+	return "", false, fmt.Errorf("no lyrics found")
 }
 
 // URL escape helper
diff --git a/main.go b/main.go
index a8f2cb5..64a0ec9 100644
--- a/main.go
+++ b/main.go
@@ -1,10 +1,13 @@
 package main
 
 import (
+	"embed"
+	"fmt"
+	"html/template"
 	"log"
 	"net/http"
 	"sync"
-	"text/template"
+	"time"
 )
 
 // version is set at build time via -ldflags="-X main.version=..."
@@ -13,62 +16,89 @@ var version = "dev"
 var importerMu sync.Mutex
 var importerRunning bool
 
-var tmpl = template.Must(template.New("index").Parse(`
-<!DOCTYPE html>
-<html>
-<head>
-	<title>Music Importer</title>
-	<style>
-		body {
-			font-family: sans-serif;
-			background: #111;
-			color: #eee;
-			text-align: center;
-			padding-top: 80px;
-		}
-		button {
-			font-size: 32px;
-			padding: 20px 40px;
-			border-radius: 10px;
-			border: none;
-			cursor: pointer;
-			background: #4CAF50;
-			color: white;
-		}
-		button:disabled {
-			background: #555;
-			cursor: not-allowed;
-		}
-		footer {
-			position: fixed;
-			bottom: 16px;
-			width: 100%;
-			font-size: 13px;
-			color: #999;
-		}
-	</style>
-</head>
-<body>
-	<h1>Music Importer</h1>
-	<form action="/run" method="POST">
-		<button type="submit" {{if .Running}}disabled{{end}}>
-			{{if .Running}}Importer Running...{{else}}Run Importer{{end}}
-		</button>
-	</form>
-	<footer>{{.Version}}</footer>
-</body>
-</html>
-`))
+//go:embed index.html.tmpl
+var tmplFS embed.FS
+
+//go:embed static
+var staticFS embed.FS
+var tmpl = template.Must(
+	template.New("index.html.tmpl").
+		Funcs(template.FuncMap{
+			// duration formats the elapsed time between two timestamps.
+			"duration": func(start, end time.Time) string {
+				if end.IsZero() {
+					return ""
+				}
+				d := end.Sub(start).Round(time.Second)
+				if d < time.Minute {
+					return d.String()
+				}
+				return fmt.Sprintf("%dm %ds", int(d.Minutes()), int(d.Seconds())%60)
+			},
+			// not is needed because Go templates have no built-in boolean negation.
+			"not": func(b bool) bool { return !b },
+			// stepCell renders a uniform step status cell.
+			// fatalStep is AlbumResult.FatalStep; when it matches the step's key
+			// the cell is marked fatal rather than a warning.
+			"stepCell": func(label string, s StepStatus, fatalStep string) template.HTML {
+				var statusClass, statusText, errHTML string
+				switch {
+				case s.Err != nil && fatalStep != "" && stepKey(label) == fatalStep:
+					statusClass = "step-fatal"
+					statusText = "✗ fatal"
+					errHTML = `<span class="step-err">` + template.HTMLEscapeString(s.Err.Error()) + `</span>`
+				case s.Err != nil:
+					statusClass = "step-warn"
+					statusText = "⚠ error"
+					errHTML = `<span class="step-err">` + template.HTMLEscapeString(s.Err.Error()) + `</span>`
+				case s.Skipped:
+					statusClass = "step-warn"
+					statusText = "– skipped"
+				default:
+					statusClass = "step-ok"
+					statusText = "✓ ok"
+				}
+				return template.HTML(`<div class="step">` +
+					`<span class="step-label">` + template.HTMLEscapeString(label) + `</span>` +
+					`<span class="` + statusClass + `">` + statusText + `</span>` +
+					errHTML +
+					`</div>`)
+			},
+		}).
+		ParseFS(tmplFS, "index.html.tmpl"),
+)
+
+// stepKey maps a human-readable step label to the FatalStep identifier used in
+// AlbumResult so the template can highlight the step that caused the abort.
+func stepKey(label string) string {
+	switch label {
+	case "Metadata":
+		return "TagMetadata"
+	case "Cover Art":
+		return "CoverArt"
+	default:
+		return label
+	}
+}
+
+type templateData struct {
+	Running bool
+	Version string
+	Session *ImportSession
+}
 
 func handleHome(w http.ResponseWriter, r *http.Request) {
 	importerMu.Lock()
 	running := importerRunning
 	importerMu.Unlock()
 
-	tmpl.Execute(w, struct {
-		Running bool
-		Version string
-	}{Running: running, Version: version})
+	if err := tmpl.Execute(w, templateData{
+		Running: running,
+		Version: version,
+		Session: lastSession,
+	}); err != nil {
+		log.Println("Template error:", err)
+	}
 }
 
 func handleRun(w http.ResponseWriter, r *http.Request) {
@@ -93,8 +123,15 @@ func handleRun(w http.ResponseWriter, r *http.Request) {
 
 func main() {
 	log.Printf("Music Importer %s starting on http://localhost:8080", version)
+	startMonitor()
+	http.Handle("/static/", http.FileServer(http.FS(staticFS)))
 	http.HandleFunc("/", handleHome)
 	http.HandleFunc("/run", handleRun)
+	http.HandleFunc("/discover/search", handleDiscoverSearch)
+	http.HandleFunc("/discover/fetch", handleDiscoverFetch)
+	http.HandleFunc("/discover/fetch/artist", handleDiscoverFetchArtist)
+	http.HandleFunc("/discover/fetch/status", handleDiscoverFetchStatus)
+	http.HandleFunc("/discover/fetch/list", handleDiscoverFetchList)
 
 	log.Fatal(http.ListenAndServe(":8080", nil))
 }
diff --git a/media.go b/media.go
index 60711a2..e0274a6 100644
--- a/media.go
+++ b/media.go
@@ -2,7 +2,11 @@ package main
 
 import (
 	"bytes"
+	"encoding/json"
 	"fmt"
+	"io"
+	"net/http"
+	"net/url"
 	"os"
 	"os/exec"
 	"path/filepath"
@@ -53,6 +57,158 @@ func EmbedAlbumArtIntoFolder(albumDir string) error {
 	return err
 }
 
+// DownloadCoverArt downloads the front cover from the Cover Art Archive and
+// saves it as cover.jpg/cover.png inside albumDir.
+// If mbid is non-empty it is used directly, bypassing the MusicBrainz search.
+// Otherwise, a search is performed using md's artist and album.
+func DownloadCoverArt(albumDir string, md *MusicMetadata, mbid string) error {
+	if mbid == "" {
+		var err error
+		mbid, err = searchMusicBrainzRelease(md.Artist, md.Album)
+		if err != nil {
+			return fmt.Errorf("MusicBrainz release search failed: %w", err)
+		}
+	}
+
+	data, ext, err := fetchCoverArtArchiveFront(mbid)
+	if err != nil {
+		return fmt.Errorf("Cover Art Archive fetch failed: %w", err)
+	}
+
+	dest := filepath.Join(albumDir, "cover."+ext)
+	if err := os.WriteFile(dest, data, 0644); err != nil {
+		return fmt.Errorf("writing cover image: %w", err)
+	}
+
+	fmt.Println("→ Downloaded cover art:", filepath.Base(dest))
+	return nil
+}
+
+// searchMusicBrainzRelease queries the MusicBrainz API for a release matching
+// the given artist and album and returns its MBID.
+func searchMusicBrainzRelease(artist, album string) (string, error) {
+	q := fmt.Sprintf(`release:"%s" AND artist:"%s"`,
+		strings.ReplaceAll(album, `"`, `\"`),
+		strings.ReplaceAll(artist, `"`, `\"`),
+	)
+	apiURL := "https://musicbrainz.org/ws/2/release/?query=" + url.QueryEscape(q) + "&fmt=json&limit=1"
+
+	req, err := http.NewRequest("GET", apiURL, nil)
+	if err != nil {
+		return "", err
+	}
+	req.Header.Set("User-Agent", "music-importer/1.0 (https://github.com/example/music-importer)")
+
+	resp, err := http.DefaultClient.Do(req)
+	if err != nil {
+		return "", err
+	}
+	defer resp.Body.Close()
+
+	if resp.StatusCode != http.StatusOK {
+		return "", fmt.Errorf("MusicBrainz returned status %d", resp.StatusCode)
+	}
+
+	var result struct {
+		Releases []struct {
+			ID string `json:"id"`
+		} `json:"releases"`
+	}
+	if err := json.NewDecoder(resp.Body).Decode(&result); err != nil {
+		return "", err
+	}
+	if len(result.Releases) == 0 {
+		return "", fmt.Errorf("no MusicBrainz release found for %q by %q", album, artist)
+	}
+	return result.Releases[0].ID, nil
+}
+
+// fetchCoverArtArchiveFront fetches the front cover image for the given
+// MusicBrainz release MBID from coverartarchive.org. It follows the 307
+// redirect to the actual image and returns the raw bytes plus the file
+// extension (e.g. "jpg" or "png").
+func fetchCoverArtArchiveFront(mbid string) ([]byte, string, error) {
+	apiURL := "https://coverartarchive.org/release/" + mbid + "/front"
+
+	resp, err := http.Get(apiURL)
+	if err != nil {
+		return nil, "", err
+	}
+	defer resp.Body.Close()
+
+	if resp.StatusCode != http.StatusOK {
+		return nil, "", fmt.Errorf("Cover Art Archive returned status %d for MBID %s", resp.StatusCode, mbid)
+	}
+
+	data, err := io.ReadAll(resp.Body)
+	if err != nil {
+		return nil, "", err
+	}
+
+	// Derive the extension from the final URL after redirect, falling back to
+	// sniffing the magic bytes.
+	ext := "jpg"
+	if finalURL := resp.Request.URL.String(); strings.HasSuffix(strings.ToLower(finalURL), ".png") {
+		ext = "png"
+	} else if bytes.HasPrefix(data, []byte{0x89, 0x50, 0x4E, 0x47}) {
+		ext = "png"
+	}
+
+	return data, ext, nil
+}
+
+const coverMaxBytes = 5 * 1024 * 1024 // 5 MB
+
+// NormalizeCoverArt checks whether the cover image in albumDir is a large
+// non-JPEG (>5 MB). If so, it converts it to JPEG and resizes it to at most
+// 2000×2000 pixels using ffmpeg, replacing the original file with cover.jpg.
+// The function is a no-op when no cover is found, the cover is already JPEG,
+// or the file is ≤5 MB.
+func NormalizeCoverArt(albumDir string) error {
+	cover, err := FindCoverImage(albumDir)
+	if err != nil {
+		return nil // no cover present, nothing to do
+	}
+
+	// Already JPEG — no conversion needed regardless of size.
+	ext := strings.ToLower(filepath.Ext(cover))
+	if ext == ".jpg" || ext == ".jpeg" {
+		return nil
+	}
+
+	info, err := os.Stat(cover)
+	if err != nil {
+		return fmt.Errorf("stat cover: %w", err)
+	}
+	if info.Size() <= coverMaxBytes {
+		return nil // small enough, leave as-is
+	}
+
+	dest := filepath.Join(albumDir, "cover.jpg")
+	fmt.Printf("→ Cover art is %.1f MB %s; converting to JPEG (max 2000×2000)…\n",
+		float64(info.Size())/(1024*1024), strings.ToUpper(strings.TrimPrefix(ext, ".")))
+
+	// scale=2000:2000:force_original_aspect_ratio=decrease fits the image within
+	// 2000×2000 while preserving aspect ratio, and never upscales smaller images.
+	cmd := exec.Command("ffmpeg", "-y", "-i", cover,
+		"-vf", "scale=2000:2000:force_original_aspect_ratio=decrease",
+		"-q:v", "2",
+		dest,
+	)
+	if out, err := cmd.CombinedOutput(); err != nil {
+		return fmt.Errorf("ffmpeg cover conversion failed: %w\n%s", err, out)
+	}
+
+	if cover != dest {
+		if err := os.Remove(cover); err != nil {
+			fmt.Println("Warning: could not remove original cover:", err)
+		}
+	}
+
+	fmt.Println("→ Converted cover art to JPEG:", filepath.Base(dest))
+	return nil
+}
+
 // -------------------------
 // Find cover image
 // -------------------------
diff --git a/metadata.go b/metadata.go
index 0861567..da146a8 100644
--- a/metadata.go
+++ b/metadata.go
@@ -1,19 +1,25 @@
 package main
 
 import (
+	"bufio"
 	"encoding/json"
 	"errors"
 	"fmt"
+	"math"
+	"os"
 	"os/exec"
 	"path/filepath"
+	"strconv"
 	"strings"
 )
 
 type MusicMetadata struct {
-	Artist string
-	Album  string
-	Title  string
-	Year   string
+	Artist  string
+	Album   string
+	Title   string
+	Year    string // four-digit year, kept for backward compat
+	Date    string // normalised as YYYY.MM.DD (or YYYY.MM or YYYY)
+	Quality string // e.g. "FLAC-24bit-96kHz" or "MP3-320kbps"
 }
 
 // Read embedded tags from an audio file using ffprobe.
@@ -39,18 +45,243 @@ func readTags(path string) (*MusicMetadata, error) {
 		return &MusicMetadata{}, nil
 	}
 
+	rawDate := firstNonEmpty(t["date"], t["DATE"], t["year"], t["YEAR"], t["ORIGINALYEAR"])
+	date := parseDate(rawDate)
+	year := ""
+	if len(date) >= 4 {
+		year = date[:4]
+	}
+
 	return &MusicMetadata{
 		Artist: firstNonEmpty(t["artist"], t["ARTIST"]),
 		Album:  firstNonEmpty(t["album"], t["ALBUM"]),
 		Title:  firstNonEmpty(t["title"], t["TITLE"]),
-		Year:   firstNonEmpty(t["year"], t["YEAR"], t["ORIGINALYEAR"]),
+		Year:   year,
+		Date:   date,
 	}, nil
 }
 
+// parseDate normalises a raw DATE/date tag value into YYYY.MM.DD (or YYYY.MM
+// or YYYY) dot-separated format, or returns the input unchanged if it cannot
+// be recognised.
+//
+// Supported input formats:
+//   - YYYY
+//   - YYYY-MM
+//   - YYYY-MM-DD
+//   - YYYYMMDD
+func parseDate(raw string) string {
+	raw = strings.TrimSpace(raw)
+	if raw == "" {
+		return ""
+	}
+
+	// YYYYMMDD (exactly 8 digits, no separators)
+	if len(raw) == 8 && isAllDigits(raw) {
+		return raw[0:4] + "." + raw[4:6] + "." + raw[6:8]
+	}
+
+	// YYYY-MM-DD, YYYY-MM, or YYYY (with dashes)
+	parts := strings.Split(raw, "-")
+	switch len(parts) {
+	case 1:
+		if len(parts[0]) == 4 && isAllDigits(parts[0]) {
+			return parts[0]
+		}
+	case 2:
+		if len(parts[0]) == 4 && isAllDigits(parts[0]) && len(parts[1]) == 2 && isAllDigits(parts[1]) {
+			return parts[0] + "." + parts[1]
+		}
+	case 3:
+		if len(parts[0]) == 4 && isAllDigits(parts[0]) &&
+			len(parts[1]) == 2 && isAllDigits(parts[1]) &&
+			len(parts[2]) == 2 && isAllDigits(parts[2]) {
+			return parts[0] + "." + parts[1] + "." + parts[2]
+		}
+	}
+
+	// Unrecognised — return as-is so we don't silently drop it.
+	return raw
+}
+
+func isAllDigits(s string) bool {
+	for _, c := range s {
+		if c < '0' || c > '9' {
+			return false
+		}
+	}
+	return len(s) > 0
+}
+
+// readAudioQuality probes the first audio stream of path and returns a
+// quality label such as "FLAC-24bit-96kHz" or "MP3-320kbps".
+func readAudioQuality(path string) (string, error) {
+	out, err := exec.Command(
+		"ffprobe", "-v", "quiet", "-print_format", "json",
+		"-show_streams", "-select_streams", "a:0",
+		path,
+	).Output()
+	if err != nil {
+		return "", err
+	}
+
+	var data struct {
+		Streams []struct {
+			CodecName        string `json:"codec_name"`
+			SampleRate       string `json:"sample_rate"`
+			BitRate          string `json:"bit_rate"`
+			BitsPerRawSample string `json:"bits_per_raw_sample"`
+		} `json:"streams"`
+	}
+
+	if err := json.Unmarshal(out, &data); err != nil {
+		return "", err
+	}
+	if len(data.Streams) == 0 {
+		return "", fmt.Errorf("no audio streams found in %s", path)
+	}
+
+	s := data.Streams[0]
+	codec := strings.ToUpper(s.CodecName) // e.g. "FLAC", "MP3"
+
+	switch strings.ToLower(s.CodecName) {
+	case "flac":
+		bits := s.BitsPerRawSample
+		if bits == "" || bits == "0" {
+			bits = "16" // safe fallback
+		}
+		khz := sampleRateToKHz(s.SampleRate)
+		return fmt.Sprintf("%s-%sbit-%s", codec, bits, khz), nil
+
+	case "mp3":
+		kbps := snapMP3Bitrate(s.BitRate)
+		return fmt.Sprintf("%s-%dkbps", codec, kbps), nil
+
+	default:
+		// Generic fallback: codec + bitrate if available.
+		if s.BitRate != "" && s.BitRate != "0" {
+			kbps := snapMP3Bitrate(s.BitRate)
+			return fmt.Sprintf("%s-%dkbps", codec, kbps), nil
+		}
+		return codec, nil
+	}
+}
+
+// sampleRateToKHz converts a sample-rate string in Hz (e.g. "44100") to a
+// human-friendly kHz string (e.g. "44.1kHz").
+func sampleRateToKHz(hz string) string {
+	n, err := strconv.Atoi(strings.TrimSpace(hz))
+	if err != nil || n == 0 {
+		return "?kHz"
+	}
+	if n%1000 == 0 {
+		return fmt.Sprintf("%dkHz", n/1000)
+	}
+	return fmt.Sprintf("%.1fkHz", float64(n)/1000.0)
+}
+
+// commonMP3Bitrates lists the standard MPEG audio bitrates in kbps.
+var commonMP3Bitrates = []int{32, 40, 48, 56, 64, 80, 96, 112, 128, 160, 192, 224, 256, 320}
+
+// snapMP3Bitrate rounds a raw bitrate string (in bps) to the nearest standard
+// MP3 bitrate (in kbps). For example "318731" → 320.
+func snapMP3Bitrate(bpsStr string) int {
+	bps, err := strconv.Atoi(strings.TrimSpace(bpsStr))
+	if err != nil || bps <= 0 {
+		return 128 // safe fallback
+	}
+	kbps := float64(bps) / 1000.0
+	best := commonMP3Bitrates[0]
+	bestDiff := math.Abs(kbps - float64(best))
+	for _, candidate := range commonMP3Bitrates[1:] {
+		if d := math.Abs(kbps - float64(candidate)); d < bestDiff {
+			bestDiff = d
+			best = candidate
+		}
+	}
+	return best
+}
+
 // Use beets to fetch metadata and tag all files in a directory.
-func tagWithBeets(path string) error {
+// A temp log file is passed to beets via -l so that skipped albums
+// (which exit 0 but produce a "skip" log entry) are detected and
+// returned as errors, triggering the MusicBrainz fallback.
+// If mbid is non-empty it is passed as --search-id to pin beets to a specific
+// MusicBrainz release. In that case, quiet mode is skipped and newlines are
+// piped to stdin so beets auto-accepts the pinned release regardless of
+// confidence score.
+func tagWithBeets(path, mbid string) error {
 	fmt.Println("→ Tagging with beets:", path)
-	return runCmd("beet", "import", "-Cq", path)
+
+	logFile, err := os.CreateTemp("", "beets-log-*.txt")
+	if err != nil {
+		return fmt.Errorf("beets: could not create temp log file: %w", err)
+	}
+	logPath := logFile.Name()
+	logFile.Close()
+	defer os.Remove(logPath)
+
+	args := []string{"import", "-C", "-l", logPath}
+	if mbid != "" {
+		// Drop -q so beets doesn't skip on low confidence. Pipe newlines to
+		// auto-accept the interactive prompt for the MBID-pinned release.
+		args = append(args, "--search-id", mbid, path)
+		cmd := exec.Command("beet", args...)
+		cmd.Stdout = os.Stdout
+		cmd.Stderr = os.Stderr
+		cmd.Stdin = strings.NewReader(strings.Repeat("A\n", 20))
+		if err := cmd.Run(); err != nil {
+			return err
+		}
+	} else {
+		args = append(args, "-q", path)
+		if err := runCmd("beet", args...); err != nil {
+			return err
+		}
+	}
+
+	// Even on exit 0, beets may have skipped the album in quiet mode.
+	// The log format is one entry per line: "<action> <path>"
+	// We treat any "skip" line as a failure so the caller falls through
+	// to the MusicBrainz lookup.
+	skipped, err := beetsLogHasSkip(logPath)
+	if err != nil {
+		// If we can't read the log, assume beets succeeded.
+		fmt.Println("beets: could not read log file:", err)
+		return nil
+	}
+	if skipped {
+		return errors.New("beets skipped album (no confident match found)")
+	}
+	return nil
+}
+
+// beetsLogHasSkip reads a beets import log file and reports whether any
+// entry has the action "skip". The log format is:
+//
+//	# beets import log
+//	<action> <path>
+//	...
+func beetsLogHasSkip(logPath string) (bool, error) {
+	f, err := os.Open(logPath)
+	if err != nil {
+		return false, err
+	}
+	defer f.Close()
+
+	scanner := bufio.NewScanner(f)
+	for scanner.Scan() {
+		line := strings.TrimSpace(scanner.Text())
+		// Skip blank lines and the header comment.
+		if line == "" || strings.HasPrefix(line, "#") {
+			continue
+		}
+		action, _, found := strings.Cut(line, " ")
+		if found && strings.EqualFold(action, "skip") {
+			return true, nil
+		}
+	}
+	return false, scanner.Err()
 }
 
 // Fallback: query MusicBrainz API manually if beets fails.
@@ -99,26 +330,44 @@ func fetchMusicBrainzInfo(filename string) (*MusicMetadata, error) {
 
 // getAlbumMetadata attempts beets tagging on the album directory, reads tags
 // back from the first track, and falls back to MusicBrainz if tags are missing.
-func getAlbumMetadata(albumPath, trackPath string) (*MusicMetadata, error) {
+// If mbid is non-empty it is forwarded to beets as --search-id.
+func getAlbumMetadata(albumPath, trackPath, mbid string) (*MusicMetadata, MetadataSource, error) {
 	fmt.Println("→ Tagging track with beets:", trackPath)
 
-	if err := tagWithBeets(albumPath); err != nil {
-		fmt.Println("Beets tagging failed; fallback to manual MusicBrainz lookup:", err)
+	beetsErr := tagWithBeets(albumPath, mbid)
+	if beetsErr != nil {
+		fmt.Println("Beets tagging failed; fallback to manual MusicBrainz lookup:", beetsErr)
 	}
 
 	md, err := readTags(trackPath)
 	if err == nil && md.Artist != "" && md.Album != "" {
-		return md, nil
+		attachQuality(md, trackPath)
+		if beetsErr == nil {
+			return md, MetadataSourceBeets, nil
+		}
+		return md, MetadataSourceFileTags, nil
 	}
 
 	fmt.Println("→ Missing tags, attempting MusicBrainz manual lookup...")
 
 	md, err = fetchMusicBrainzInfo(trackPath)
 	if err != nil {
-		return nil, fmt.Errorf("metadata lookup failed: %w", err)
+		return nil, MetadataSourceUnknown, fmt.Errorf("metadata lookup failed: %w", err)
 	}
 
-	return md, nil
+	attachQuality(md, trackPath)
+	return md, MetadataSourceMusicBrainz, nil
+}
+
+// attachQuality probes trackPath for audio quality and sets md.Quality.
+// Errors are logged but not returned — a missing quality label is non-fatal.
+func attachQuality(md *MusicMetadata, trackPath string) {
+	q, err := readAudioQuality(trackPath)
+	if err != nil {
+		fmt.Println("Could not determine audio quality:", err)
+		return
+	}
+	md.Quality = q
 }
 
 func firstNonEmpty(vals ...string) string {
diff --git a/monitor.go b/monitor.go
new file mode 100644
index 0000000..c09438b
--- /dev/null
+++ b/monitor.go
@@ -0,0 +1,304 @@
+package main
+
+import (
+	"fmt"
+	"log"
+	"os"
+	"path/filepath"
+	"strings"
+	"sync"
+	"time"
+)
+
+// pendingDownload tracks a queued slskd download that should be auto-imported
+// once all files have transferred successfully.
+type pendingDownload struct {
+	ID        string      // dedup key (release MBID for single fetches; release group MBID for artist fetches)
+	BeetsMBID string      // release MBID passed to beets --search-id (may differ from ID)
+	Artist    string
+	Album     string
+	Username  string      // slskd peer username
+	Dir       string      // remote directory path on the peer
+	Files     []slskdFile // files that were queued for download
+	Entry     *fetchEntry // fetch card to update with import progress
+	TrackCount int        // expected number of audio tracks (0 = unknown, skip check)
+}
+
+var (
+	pendingMu        sync.Mutex
+	pendingDownloads = make(map[string]*pendingDownload) // keyed by MBID
+)
+
+// registerDownload records a queued slskd download for monitoring and eventual
+// auto-import. id is used as the dedup key; beetsMBID is the release MBID
+// forwarded to beets --search-id (may be empty or differ from id).
+// trackCount is the expected number of audio tracks from MusicBrainz; 0 means
+// unknown and the sanity check will be skipped at import time.
+// If entry is nil a new fetchEntry is created so the frontend can discover it
+// via /discover/fetch/list.
+func registerDownload(id, beetsMBID, artist, album string, trackCount int, folder *albumFolder, entry *fetchEntry) {
+	pd := &pendingDownload{
+		ID:         id,
+		BeetsMBID:  beetsMBID,
+		Artist:     artist,
+		Album:      album,
+		Username:   folder.Username,
+		Dir:        folder.Dir,
+		Files:      folder.Files,
+		Entry:      entry,
+		TrackCount: trackCount,
+	}
+
+	if entry == nil {
+		e := newFetchEntry(id, artist, album)
+		e.appendLog(fmt.Sprintf("Queued %d files from %s — waiting for download",
+			len(folder.Files), folder.Username))
+		pd.Entry = e
+	}
+
+	pendingMu.Lock()
+	pendingDownloads[id] = pd
+	pendingMu.Unlock()
+
+	log.Printf("[monitor] registered: %q by %s (id: %s, beets mbid: %s, peer: %s, %d files, expected tracks: %d)",
+		album, artist, id, beetsMBID, folder.Username, len(folder.Files), trackCount)
+}
+
+// startMonitor launches a background goroutine that periodically checks whether
+// pending downloads have completed and triggers import when they have.
+func startMonitor() {
+	go func() {
+		for {
+			time.Sleep(15 * time.Second)
+			checkPendingDownloads()
+		}
+	}()
+	log.Println("[monitor] started")
+}
+
+// checkPendingDownloads polls slskd transfer state for every pending download
+// and kicks off importPendingRelease for any that are fully complete.
+func checkPendingDownloads() {
+	pendingMu.Lock()
+	if len(pendingDownloads) == 0 {
+		pendingMu.Unlock()
+		return
+	}
+	snapshot := make(map[string]*pendingDownload, len(pendingDownloads))
+	for k, v := range pendingDownloads {
+		snapshot[k] = v
+	}
+	pendingMu.Unlock()
+
+	log.Printf("[monitor] checking %d pending download(s)", len(snapshot))
+
+	// Group by username to minimise API calls.
+	byUser := make(map[string][]*pendingDownload)
+	for _, pd := range snapshot {
+		byUser[pd.Username] = append(byUser[pd.Username], pd)
+	}
+
+	for username, pds := range byUser {
+		dirs, err := getSlskdTransfers(username)
+		if err != nil {
+			log.Printf("[monitor] failed to get transfers for %s: %v", username, err)
+			continue
+		}
+
+		// Index transfer dirs by normalised path.
+		transfersByDir := make(map[string][]slskdTransferFile, len(dirs))
+		for _, d := range dirs {
+			norm := strings.ReplaceAll(d.Directory, "\\", "/")
+			transfersByDir[norm] = d.Files
+		}
+
+		for _, pd := range pds {
+			normDir := strings.ReplaceAll(pd.Dir, "\\", "/")
+			files, ok := transfersByDir[normDir]
+			if !ok {
+				log.Printf("[monitor] transfer dir not found yet for %q (peer: %s)", pd.Dir, username)
+				continue
+			}
+
+			if !allFilesCompleted(files) {
+				log.Printf("[monitor] %q by %s: download still in progress", pd.Album, pd.Artist)
+				continue
+			}
+
+			localDir := localDirForDownload(pd, files)
+			if localDir == "" {
+				log.Printf("[monitor] cannot determine local dir for %q by %s", pd.Album, pd.Artist)
+				pd.Entry.appendLog("Error: could not determine local download path from transfer info")
+				continue
+			}
+
+			log.Printf("[monitor] download complete: %q by %s → %s", pd.Album, pd.Artist, localDir)
+
+			// Remove from pending before starting import to avoid double-import.
+			pendingMu.Lock()
+			delete(pendingDownloads, pd.ID)
+			pendingMu.Unlock()
+
+			go importPendingRelease(pd, localDir)
+		}
+	}
+}
+
+// allFilesCompleted reports whether every file in a transfer directory has
+// reached a terminal Completed state. Returns false if files is empty.
+func allFilesCompleted(files []slskdTransferFile) bool {
+	if len(files) == 0 {
+		return false
+	}
+	for _, f := range files {
+		if !strings.Contains(f.State, "Completed") {
+			return false
+		}
+	}
+	return true
+}
+
+// localDirForDownload resolves the local filesystem path for a completed download.
+//
+// Strategy 1 — localFilename from transfer metadata: slskd sets this field to
+// the absolute path of the downloaded file. Works when paths are consistent
+// across containers (same volume mount point).
+//
+// Strategy 2 — SLSKD_DOWNLOAD_DIR reconstruction: slskd stores files under
+// {downloadDir}/{username}/{sanitized_remote_dir}/. Used when localFilename is
+// empty or when SLSKD_DOWNLOAD_DIR is explicitly set to override.
+func localDirForDownload(pd *pendingDownload, files []slskdTransferFile) string {
+	// Strategy 1: use localFilename from transfer response.
+	for _, f := range files {
+		if f.LocalFilename != "" {
+			dir := filepath.Dir(f.LocalFilename)
+			log.Printf("[monitor] local dir from localFilename: %s", dir)
+			return dir
+		}
+	}
+
+	// Strategy 2: reconstruct from SLSKD_DOWNLOAD_DIR.
+	// slskd places downloaded files directly into {downloadDir}/{album_folder_name}/,
+	// where the folder name is the last path component of the remote directory.
+	dlDir := os.Getenv("SLSKD_DOWNLOAD_DIR")
+	if dlDir == "" {
+		log.Printf("[monitor] localFilename empty and SLSKD_DOWNLOAD_DIR not set — cannot determine local dir for %q", pd.Album)
+		return ""
+	}
+
+	dir := filepath.Join(dlDir, filepath.Base(filepath.FromSlash(pd.Dir)))
+	log.Printf("[monitor] local dir reconstructed from SLSKD_DOWNLOAD_DIR: %s", dir)
+	return dir
+}
+
+// importPendingRelease runs the full import pipeline on a completed download.
+// It mirrors RunImporter's per-album logic but uses the MBID for beets tagging.
+func importPendingRelease(pd *pendingDownload, localDir string) {
+	entry := pd.Entry
+	logf := func(msg string) {
+		entry.appendLog("[import] " + msg)
+		log.Printf("[monitor/import %s] %s", pd.ID, msg)
+	}
+
+	logf(fmt.Sprintf("Starting import from %s", localDir))
+
+	libraryDir := os.Getenv("LIBRARY_DIR")
+	if libraryDir == "" {
+		entry.finish(fmt.Errorf("LIBRARY_DIR is not set"))
+		return
+	}
+
+	tracks, err := getAudioFiles(localDir)
+	if err != nil {
+		entry.finish(fmt.Errorf("scanning audio files: %w", err))
+		return
+	}
+	if len(tracks) == 0 {
+		entry.finish(fmt.Errorf("no audio files found in %s", localDir))
+		return
+	}
+	logf(fmt.Sprintf("Found %d tracks", len(tracks)))
+
+	if pd.TrackCount > 0 && len(tracks) != pd.TrackCount {
+		entry.finish(fmt.Errorf(
+			"track count mismatch: downloaded %d tracks but release expects %d — aborting to avoid importing wrong edition",
+			len(tracks), pd.TrackCount,
+		))
+		return
+	}
+
+	if err := cleanAlbumTags(localDir); err != nil {
+		logf(fmt.Sprintf("Clean tags warning: %v", err))
+	}
+
+	md, src, err := getAlbumMetadata(localDir, tracks[0], pd.BeetsMBID)
+	if err != nil {
+		entry.finish(fmt.Errorf("metadata failed: %w", err))
+		return
+	}
+	logf(fmt.Sprintf("Tagged via %s: %s — %s", src, md.Artist, md.Album))
+
+	if _, err := DownloadAlbumLyrics(localDir); err != nil {
+		logf(fmt.Sprintf("Lyrics warning: %v", err))
+	}
+
+	if err := applyReplayGain(localDir); err != nil {
+		entry.finish(fmt.Errorf("ReplayGain failed: %w", err))
+		return
+	}
+	logf("ReplayGain applied")
+
+	if _, err := FindCoverImage(localDir); err != nil {
+		if err := DownloadCoverArt(localDir, md, pd.BeetsMBID); err != nil {
+			logf(fmt.Sprintf("Cover art download warning: %v", err))
+		}
+	}
+
+	if err := NormalizeCoverArt(localDir); err != nil {
+		logf(fmt.Sprintf("Cover art normalization warning: %v", err))
+	}
+
+	if err := EmbedAlbumArtIntoFolder(localDir); err != nil {
+		entry.finish(fmt.Errorf("cover embed failed: %w", err))
+		return
+	}
+	logf("Cover art embedded")
+
+	targetDir := albumTargetDir(libraryDir, md)
+	if _, err := os.Stat(targetDir); err == nil {
+		logf(fmt.Sprintf("Album already exists in library, skipping move: %s", targetDir))
+		entry.finish(nil)
+		return
+	}
+
+	var moveErr error
+	for _, track := range tracks {
+		if err := moveToLibrary(libraryDir, md, track); err != nil {
+			logf(fmt.Sprintf("Move warning: %v", err))
+			moveErr = err
+		}
+	}
+
+	lyrics, _ := getLyricFiles(localDir)
+	for _, file := range lyrics {
+		if err := moveToLibrary(libraryDir, md, file); err != nil {
+			logf(fmt.Sprintf("Move lyrics warning: %v", err))
+		}
+	}
+
+	if coverImg, err := FindCoverImage(localDir); err == nil {
+		if err := moveToLibrary(libraryDir, md, coverImg); err != nil {
+			logf(fmt.Sprintf("Move cover warning: %v", err))
+		}
+	}
+
+	os.Remove(localDir)
+
+	if moveErr != nil {
+		entry.finish(fmt.Errorf("import completed with move errors: %w", moveErr))
+		return
+	}
+
+	logf("Import complete")
+	entry.finish(nil)
+}
diff --git a/slskd.go b/slskd.go
new file mode 100644
index 0000000..70fcad9
--- /dev/null
+++ b/slskd.go
@@ -0,0 +1,460 @@
+package main
+
+import (
+	"bytes"
+	"encoding/json"
+	"fmt"
+	"io"
+	"log"
+	"net/http"
+	"os"
+	"path"
+	"strings"
+	"time"
+)
+
+// slskdAttr is a Soulseek file attribute (bitrate, sample rate, bit depth, etc.).
+// Attribute types: 0 = bitrate (kbps), 1 = duration (s), 2 = VBR flag,
+//
+//	4 = sample rate (Hz), 5 = bit depth.
+type slskdAttr struct {
+	Type  int `json:"type"`
+	Value int `json:"value"`
+}
+
+// slskdFile is a single file in a slskd search response.
+type slskdFile struct {
+	Filename   string      `json:"filename"`
+	Size       int64       `json:"size"`
+	Extension  string      `json:"extension"`
+	Attributes []slskdAttr `json:"attributes"`
+}
+
+// slskdPeerResponse is one peer's response to a search.
+type slskdPeerResponse struct {
+	Username string      `json:"username"`
+	Files    []slskdFile `json:"files"`
+}
+
+// slskdSearch is the search-state object returned by GET /api/v0/searches/{id}.
+// File responses are not included here; fetch them from /searches/{id}/responses.
+type slskdSearch struct {
+	ID    string `json:"id"`
+	State string `json:"state"`
+}
+
+// Quality tiers; higher value = more preferred.
+const (
+	qualityUnknown   = 0
+	qualityMP3Any    = 1
+	qualityMP3_320   = 2
+	qualityFLACOther = 3 // FLAC at unspecified or uncommon specs
+	qualityFLAC24_96 = 4
+	qualityFLAC16_44 = 5 // most preferred: standard CD-quality lossless
+)
+
+// albumFolder groups audio files from the same peer and directory path.
+type albumFolder struct {
+	Username string
+	Dir      string
+	Files    []slskdFile
+	Quality  int
+}
+
+func slskdBaseURL() string {
+	return strings.TrimRight(os.Getenv("SLSKD_URL"), "/")
+}
+
+// slskdDo performs an authenticated HTTP request against the slskd API.
+func slskdDo(method, endpoint string, body interface{}) (*http.Response, error) {
+	base := slskdBaseURL()
+	if base == "" {
+		return nil, fmt.Errorf("SLSKD_URL is not configured")
+	}
+
+	var br io.Reader
+	if body != nil {
+		data, err := json.Marshal(body)
+		if err != nil {
+			return nil, err
+		}
+		br = bytes.NewReader(data)
+	}
+
+	req, err := http.NewRequest(method, base+endpoint, br)
+	if err != nil {
+		return nil, err
+	}
+	if key := os.Getenv("SLSKD_API_KEY"); key != "" {
+		req.Header.Set("X-API-Key", key)
+	}
+	if body != nil {
+		req.Header.Set("Content-Type", "application/json")
+	}
+
+	return http.DefaultClient.Do(req)
+}
+
+// createSlskdSearch starts a new slskd search and returns its ID.
+func createSlskdSearch(searchText string) (string, error) {
+	payload := map[string]interface{}{
+		"searchText":             searchText,
+		"fileLimit":              10000,
+		"filterResponses":        true,
+		"maximumPeerQueueLength": 1000,
+		"minimumPeerUploadSpeed": 0,
+		"responseLimit":          100,
+		"timeout":                15000,
+	}
+
+	resp, err := slskdDo("POST", "/api/v0/searches", payload)
+	if err != nil {
+		return "", err
+	}
+	defer resp.Body.Close()
+
+	if resp.StatusCode != http.StatusCreated && resp.StatusCode != http.StatusOK {
+		b, _ := io.ReadAll(resp.Body)
+		return "", fmt.Errorf("slskd search failed (%d): %s", resp.StatusCode, strings.TrimSpace(string(b)))
+	}
+
+	var s slskdSearch
+	if err := json.NewDecoder(resp.Body).Decode(&s); err != nil {
+		return "", err
+	}
+	return s.ID, nil
+}
+
+// slskdSearchIsTerminal reports whether a slskd SearchStates string has reached
+// a terminal state. slskd serialises its [Flags] enum as a comma-separated list
+// (e.g. "Completed, TimedOut"), so we check for containment rather than equality.
+func slskdSearchIsTerminal(state string) bool {
+	for _, term := range []string{"Completed", "TimedOut", "Errored", "Cancelled"} {
+		if strings.Contains(state, term) {
+			return true
+		}
+	}
+	return false
+}
+
+// pollSlskdSearch waits up to 30 s for a search to reach a terminal state,
+// then returns the responses from the dedicated /responses sub-endpoint.
+// Each poll check-in is reported via logf.
+func pollSlskdSearch(id string, logf func(string)) ([]slskdPeerResponse, error) {
+	deadline := time.Now().Add(60 * time.Second)
+	for {
+		resp, err := slskdDo("GET", "/api/v0/searches/"+id, nil)
+		if err != nil {
+			return nil, err
+		}
+		var s slskdSearch
+		err = json.NewDecoder(resp.Body).Decode(&s)
+		resp.Body.Close()
+		if err != nil {
+			return nil, err
+		}
+
+		logf(fmt.Sprintf("Search state: %s", s.State))
+
+		if slskdSearchIsTerminal(s.State) {
+			return fetchSlskdResponses(id, logf)
+		}
+
+		if time.Now().After(deadline) {
+			logf("Poll deadline reached, fetching current results")
+			return fetchSlskdResponses(id, logf)
+		}
+		time.Sleep(2 * time.Second)
+	}
+}
+
+// fetchSlskdResponses fetches file responses from the dedicated sub-endpoint.
+// The main GET /searches/{id} endpoint only returns metadata; responses live at
+// /searches/{id}/responses.
+func fetchSlskdResponses(id string, logf func(string)) ([]slskdPeerResponse, error) {
+	resp, err := slskdDo("GET", "/api/v0/searches/"+id+"/responses", nil)
+	if err != nil {
+		return nil, err
+	}
+	defer resp.Body.Close()
+
+	if resp.StatusCode != http.StatusOK {
+		b, _ := io.ReadAll(resp.Body)
+		return nil, fmt.Errorf("fetching responses failed (%d): %s", resp.StatusCode, strings.TrimSpace(string(b)))
+	}
+
+	var responses []slskdPeerResponse
+	if err := json.NewDecoder(resp.Body).Decode(&responses); err != nil {
+		return nil, fmt.Errorf("decoding responses: %w", err)
+	}
+	logf(fmt.Sprintf("Fetched %d peer responses", len(responses)))
+	return responses, nil
+}
+
+// deleteSlskdSearch removes a search from slskd (best-effort cleanup).
+func deleteSlskdSearch(id string) {
+	resp, err := slskdDo("DELETE", "/api/v0/searches/"+id, nil)
+	if err == nil {
+		resp.Body.Close()
+	}
+}
+
+// fileDir returns the directory portion of a Soulseek filename,
+// normalising backslashes to forward slashes first.
+func fileDir(filename string) string {
+	return path.Dir(strings.ReplaceAll(filename, "\\", "/"))
+}
+
+// normaliseExt returns a lower-case extension that always starts with ".".
+func normaliseExt(raw string) string {
+	ext := strings.ToLower(raw)
+	if ext != "" && !strings.HasPrefix(ext, ".") {
+		ext = "." + ext
+	}
+	return ext
+}
+
+// fileQuality scores a single file by the preferred quality tier.
+func fileQuality(f slskdFile) int {
+	ext := normaliseExt(f.Extension)
+	if ext == "." || ext == "" {
+		ext = strings.ToLower(path.Ext(strings.ReplaceAll(f.Filename, "\\", "/")))
+	}
+
+	switch ext {
+	case ".flac":
+		var depth, rate int
+		for _, a := range f.Attributes {
+			switch a.Type {
+			case 4:
+				rate = a.Value
+			case 5:
+				depth = a.Value
+			}
+		}
+		if depth == 16 && rate == 44100 {
+			return qualityFLAC16_44
+		}
+		if depth == 24 && rate == 96000 {
+			return qualityFLAC24_96
+		}
+		return qualityFLACOther
+
+	case ".mp3":
+		for _, a := range f.Attributes {
+			if a.Type == 0 && a.Value >= 315 {
+				return qualityMP3_320
+			}
+		}
+		return qualityMP3Any
+	}
+
+	return qualityUnknown
+}
+
+// groupAlbumFolders groups audio files by (username, directory) and scores each group.
+func groupAlbumFolders(responses []slskdPeerResponse) []albumFolder {
+	type key struct{ user, dir string }
+	m := make(map[key]*albumFolder)
+
+	for _, r := range responses {
+		for _, f := range r.Files {
+			ext := normaliseExt(f.Extension)
+			if ext == "." || ext == "" {
+				ext = strings.ToLower(path.Ext(strings.ReplaceAll(f.Filename, "\\", "/")))
+			}
+			if ext != ".flac" && ext != ".mp3" {
+				continue
+			}
+
+			k := key{r.Username, fileDir(f.Filename)}
+			if m[k] == nil {
+				m[k] = &albumFolder{Username: r.Username, Dir: k.dir}
+			}
+			m[k].Files = append(m[k].Files, f)
+			if q := fileQuality(f); q > m[k].Quality {
+				m[k].Quality = q
+			}
+		}
+	}
+
+	out := make([]albumFolder, 0, len(m))
+	for _, af := range m {
+		out = append(out, *af)
+	}
+	return out
+}
+
+// bestAlbumFolder picks the highest-quality folder; file count breaks ties.
+func bestAlbumFolder(folders []albumFolder) *albumFolder {
+	if len(folders) == 0 {
+		return nil
+	}
+	best := &folders[0]
+	for i := 1; i < len(folders); i++ {
+		a := &folders[i]
+		if a.Quality > best.Quality || (a.Quality == best.Quality && len(a.Files) > len(best.Files)) {
+			best = a
+		}
+	}
+	return best
+}
+
+// queueSlskdDownload sends a batch download request to slskd for all files in folder.
+func queueSlskdDownload(folder *albumFolder) error {
+	type dlFile struct {
+		Filename string `json:"filename"`
+		Size     int64  `json:"size"`
+	}
+	files := make([]dlFile, len(folder.Files))
+	for i, f := range folder.Files {
+		files[i] = dlFile{Filename: f.Filename, Size: f.Size}
+	}
+
+	resp, err := slskdDo("POST", "/api/v0/transfers/downloads/"+folder.Username, files)
+	if err != nil {
+		return err
+	}
+	defer resp.Body.Close()
+
+	if resp.StatusCode != http.StatusCreated && resp.StatusCode != http.StatusOK {
+		b, _ := io.ReadAll(resp.Body)
+		return fmt.Errorf("slskd download request failed (%d): %s", resp.StatusCode, strings.TrimSpace(string(b)))
+	}
+	return nil
+}
+
+// qualityLabel returns a human-readable label for a quality tier constant.
+func qualityLabel(q int) string {
+	switch q {
+	case qualityFLAC16_44:
+		return "FLAC 16bit/44.1kHz"
+	case qualityFLAC24_96:
+		return "FLAC 24bit/96kHz"
+	case qualityFLACOther:
+		return "FLAC"
+	case qualityMP3_320:
+		return "MP3 320kbps"
+	case qualityMP3Any:
+		return "MP3"
+	default:
+		return "unknown"
+	}
+}
+
+// slskdTransferFile is one file entry in a slskd transfers response.
+type slskdTransferFile struct {
+	Filename      string `json:"filename"`
+	LocalFilename string `json:"localFilename"`
+	State         string `json:"state"`
+	Size          int64  `json:"size"`
+}
+
+// slskdTransferDir groups transfer files by remote directory.
+type slskdTransferDir struct {
+	Directory string              `json:"directory"`
+	Files     []slskdTransferFile `json:"files"`
+}
+
+// slskdUserTransfers is the object returned by GET /api/v0/transfers/downloads/{username}.
+type slskdUserTransfers struct {
+	Directories []slskdTransferDir `json:"directories"`
+}
+
+// getSlskdTransfers returns all active/pending download transfer directories for a peer.
+func getSlskdTransfers(username string) ([]slskdTransferDir, error) {
+	resp, err := slskdDo("GET", "/api/v0/transfers/downloads/"+username, nil)
+	if err != nil {
+		return nil, err
+	}
+	defer resp.Body.Close()
+
+	if resp.StatusCode != http.StatusOK {
+		b, _ := io.ReadAll(resp.Body)
+		return nil, fmt.Errorf("slskd transfers (%d): %s", resp.StatusCode, strings.TrimSpace(string(b)))
+	}
+
+	var ut slskdUserTransfers
+	if err := json.NewDecoder(resp.Body).Decode(&ut); err != nil {
+		return nil, err
+	}
+	return ut.Directories, nil
+}
+
+// fetchRelease searches slskd for an album, queues the best-quality match for
+// download, and returns the chosen folder so the caller can monitor completion.
+// mbid, if non-empty, will be stored for use during import (beets --search-id).
+// trackCount, if > 0, filters candidate folders to those whose audio file count
+// matches the expected number of tracks on the release, so alternate editions
+// with different track counts are not accidentally selected.
+func fetchRelease(artist, album, mbid string, trackCount int, logf func(string)) (*albumFolder, error) {
+	query := artist + " " + album
+	log.Printf("[discover] fetch started: %q by %s (expected tracks: %d)", album, artist, trackCount)
+	logf("Starting fetch for: " + query)
+
+	logf("Creating slskd search…")
+	id, err := createSlskdSearch(query)
+	if err != nil {
+		return nil, fmt.Errorf("create search: %w", err)
+	}
+	log.Printf("[discover] slskd search created: %s", id)
+	logf(fmt.Sprintf("Search created (id: %s)", id))
+	defer func() {
+		log.Printf("[discover] deleting slskd search %s", id)
+		deleteSlskdSearch(id)
+	}()
+
+	logf("Polling for results…")
+	responses, err := pollSlskdSearch(id, logf)
+	if err != nil {
+		return nil, fmt.Errorf("poll search: %w", err)
+	}
+	log.Printf("[discover] search %s finished: %d peer responses", id, len(responses))
+	logf(fmt.Sprintf("Search finished: %d peer responses received", len(responses)))
+
+	logf("Grouping results into album folders…")
+	folders := groupAlbumFolders(responses)
+	log.Printf("[discover] grouped into %d candidate album folders", len(folders))
+	logf(fmt.Sprintf("Found %d candidate album folders", len(folders)))
+
+	if len(folders) == 0 {
+		return nil, fmt.Errorf("no audio files found for %q by %s", album, artist)
+	}
+
+	// When we know the expected track count, prefer folders that match exactly
+	// so we don't accidentally grab a bonus-track edition or a different version
+	// that won't align with the release MBID we pass to beets.
+	candidates := folders
+	if trackCount > 0 {
+		var matched []albumFolder
+		for _, f := range folders {
+			if len(f.Files) == trackCount {
+				matched = append(matched, f)
+			}
+		}
+		if len(matched) > 0 {
+			log.Printf("[discover] %d/%d folders match expected track count (%d)", len(matched), len(folders), trackCount)
+			logf(fmt.Sprintf("Filtered to %d/%d folders matching expected track count (%d)",
+				len(matched), len(folders), trackCount))
+			candidates = matched
+		} else {
+			log.Printf("[discover] no folders matched expected track count (%d); using best available", trackCount)
+			logf(fmt.Sprintf("Warning: no folders matched expected track count (%d); using best available", trackCount))
+		}
+	}
+
+	best := bestAlbumFolder(candidates)
+	log.Printf("[discover] selected folder: %s from %s (%s, %d files)",
+		best.Dir, best.Username, qualityLabel(best.Quality), len(best.Files))
+	logf(fmt.Sprintf("Selected folder: %s", best.Dir))
+	logf(fmt.Sprintf("  Peer: %s | Quality: %s | Files: %d",
+		best.Username, qualityLabel(best.Quality), len(best.Files)))
+
+	logf(fmt.Sprintf("Queuing %d files for download…", len(best.Files)))
+	if err := queueSlskdDownload(best); err != nil {
+		return nil, fmt.Errorf("queue download: %w", err)
+	}
+	log.Printf("[discover] download queued: %d files from %s", len(best.Files), best.Username)
+	logf("Download queued — waiting for completion before import")
+	return best, nil
+}
diff --git a/static/app.js b/static/app.js
new file mode 100644
index 0000000..e4e4104
--- /dev/null
+++ b/static/app.js
@@ -0,0 +1,313 @@
+"use strict";
+
+// IDs of fetch cards we've already created, so we don't duplicate them.
+const knownFetchIds = new Set();
+
+document.addEventListener("DOMContentLoaded", () => {
+  initTabs();
+  initSearch();
+  initFetchList();
+});
+
+// ── Tabs ───────────────────────────────────────────────────────────────────────
+
+function initTabs() {
+  document.querySelector(".tabs").addEventListener("click", (e) => {
+    const btn = e.target.closest(".tab-btn");
+    if (!btn) return;
+    showTab(btn.dataset.tab);
+  });
+}
+
+function showTab(name) {
+  document
+    .querySelectorAll(".tab-pane")
+    .forEach((p) => p.classList.remove("active"));
+  document
+    .querySelectorAll(".tab-btn")
+    .forEach((b) => b.classList.remove("active"));
+  document.getElementById("tab-" + name).classList.add("active");
+  document
+    .querySelector(`.tab-btn[data-tab="${name}"]`)
+    .classList.add("active");
+}
+
+// ── Search ─────────────────────────────────────────────────────────────────────
+
+let searchType = "release";
+
+function initSearch() {
+  document.querySelector(".type-toggle").addEventListener("click", (e) => {
+    const btn = e.target.closest(".type-btn");
+    if (btn) setSearchType(btn.dataset.type);
+  });
+
+  const searchBtn = document.getElementById("search-btn");
+  const searchInput = document.getElementById("search-q");
+  searchBtn.addEventListener("click", doSearch);
+  searchInput.addEventListener("keydown", (e) => {
+    if (e.key === "Enter") doSearch();
+  });
+
+  // Event delegation for dynamically rendered result buttons
+  document.getElementById("search-results").addEventListener("click", (e) => {
+    const btn = e.target.closest(".fetch-btn");
+    if (!btn || btn.disabled) return;
+    if (btn.dataset.fetchType === "artist") startArtistFetch(btn);
+    else startReleaseFetch(btn);
+  });
+}
+
+function setSearchType(type) {
+  searchType = type;
+  document.querySelectorAll(".type-btn").forEach((b) => {
+    b.classList.toggle("active", b.dataset.type === type);
+  });
+}
+
+function doSearch() {
+  const q = document.getElementById("search-q").value.trim();
+  if (!q) return;
+
+  const btn = document.getElementById("search-btn");
+  const resultsEl = document.getElementById("search-results");
+
+  btn.disabled = true;
+  btn.textContent = "Searching\u2026";
+  resultsEl.innerHTML = '<p class="search-msg">Searching MusicBrainz\u2026</p>';
+
+  fetch(`/discover/search?q=${encodeURIComponent(q)}&type=${searchType}`)
+    .then((r) => {
+      if (!r.ok)
+        return r.text().then((t) => {
+          throw new Error(t || r.statusText);
+        });
+      return r.json();
+    })
+    .then((data) => renderResults(data))
+    .catch((err) => {
+      resultsEl.innerHTML = `<p class="search-msg error">Error: ${esc(err.message)}</p>`;
+    })
+    .finally(() => {
+      btn.disabled = false;
+      btn.textContent = "Search";
+    });
+}
+
+// ── Results rendering ──────────────────────────────────────────────────────────
+
+function renderResults(data) {
+  const el = document.getElementById("search-results");
+  if (!data || data.length === 0) {
+    el.innerHTML = '<p class="search-msg">No results found.</p>';
+    return;
+  }
+  const renderer = searchType === "artist" ? renderArtist : renderRelease;
+  el.innerHTML = data.map(renderer).join("");
+}
+
+function renderRelease(r) {
+  const credits = r["artist-credit"] ?? [];
+  const artist =
+    credits.map((c) => c.name || c.artist?.name || "").join("") ||
+    "Unknown Artist";
+  const year = r.date?.substring(0, 4) ?? "";
+  const type = r["release-group"]?.["primary-type"] ?? "";
+  const country = r.country ?? "";
+  const formats = [
+    ...new Set((r.media ?? []).map((m) => m.format).filter(Boolean)),
+  ].join("+");
+  const lang = r["text-representation"]?.language ?? "";
+  const meta = [year, type, formats, country, lang]
+    .filter(Boolean)
+    .join(" \u00b7 ");
+  const dis = r.disambiguation ? ` (${esc(r.disambiguation)})` : "";
+  const coverUrl = `https://coverartarchive.org/release/${r.id}/front-250`;
+
+  return `
+    <div class="result-row">
+      <img class="result-cover" src="${coverUrl}" onerror="this.style.display='none'" loading="lazy" alt="">
+      <div class="result-info">
+        <span class="result-title">${esc(artist)} \u2014 ${esc(r.title)}<span class="result-dis">${dis}</span></span>
+        ${meta ? `<span class="result-meta">${esc(meta)}</span>` : ""}
+      </div>
+      <button class="fetch-btn"
+        data-fetch-type="release"
+        data-id="${esc(r.id)}"
+        data-artist="${esc(artist)}"
+        data-album="${esc(r.title)}">Fetch</button>
+    </div>`;
+}
+
+function renderArtist(a) {
+  const dis = a.disambiguation ? ` (${esc(a.disambiguation)})` : "";
+  return `
+    <div class="result-row">
+      <div class="result-info">
+        <span class="result-title">${esc(a.name)}${dis}</span>
+        ${a.country ? `<span class="result-meta">${esc(a.country)}</span>` : ""}
+      </div>
+      <button class="fetch-btn"
+        data-fetch-type="artist"
+        data-id="${esc(a.id)}"
+        data-name="${esc(a.name)}">Fetch All</button>
+    </div>`;
+}
+
+// ── Fetch operations ───────────────────────────────────────────────────────────
+
+function startReleaseFetch(btn) {
+  const { id, artist, album } = btn.dataset;
+  btn.disabled = true;
+  btn.textContent = "Fetching\u2026";
+
+  fetch("/discover/fetch", {
+    method: "POST",
+    headers: { "Content-Type": "application/json" },
+    body: JSON.stringify({ id, artist, album }),
+  })
+    .then((r) => {
+      if (!r.ok)
+        return r.text().then((t) => {
+          throw new Error(t || r.statusText);
+        });
+      return r.json();
+    })
+    .then(() => {
+      addFetchCard(id, `${artist} \u2014 ${album}`);
+      pollFetch(id);
+    })
+    .catch((err) => {
+      btn.disabled = false;
+      btn.textContent = "Fetch";
+      showFetchError(err.message);
+    });
+}
+
+function startArtistFetch(btn) {
+  const { id, name } = btn.dataset;
+  btn.disabled = true;
+  btn.textContent = "Fetching\u2026";
+
+  fetch("/discover/fetch/artist", {
+    method: "POST",
+    headers: { "Content-Type": "application/json" },
+    body: JSON.stringify({ id, name }),
+  })
+    .then((r) => {
+      if (!r.ok)
+        return r.text().then((t) => {
+          throw new Error(t || r.statusText);
+        });
+      return r.json();
+    })
+    .then(() => {
+      addFetchCard(id, `${name} \u2014 full discography`);
+      pollFetch(id);
+    })
+    .catch((err) => {
+      btn.disabled = false;
+      btn.textContent = "Fetch All";
+      showFetchError(err.message);
+    });
+}
+
+// ── Fetch cards ────────────────────────────────────────────────────────────────
+
+function addFetchCard(id, title) {
+  knownFetchIds.add(id);
+  const list = document.getElementById("fetch-list");
+  const card = document.createElement("div");
+  card.className = "fetch-card";
+  card.id = `fetch-${id}`;
+  card.innerHTML = `
+    <div class="fetch-header">
+      <span class="fetch-title">${esc(title)}</span>
+      <span class="fetch-status" id="fstatus-${id}">In progress\u2026</span>
+    </div>
+    <div class="fetch-log" id="flog-${id}"></div>`;
+  list.prepend(card);
+}
+
+function pollFetch(id) {
+  fetch(`/discover/fetch/status?id=${encodeURIComponent(id)}`)
+    .then((r) => r.json())
+    .then((data) => {
+      const logEl = document.getElementById(`flog-${id}`);
+      const statusEl = document.getElementById(`fstatus-${id}`);
+      const card = document.getElementById(`fetch-${id}`);
+
+      if (logEl && data.log) {
+        logEl.innerHTML = data.log
+          .map((l) => `<div class="log-line">${esc(l)}</div>`)
+          .join("");
+        logEl.scrollTop = logEl.scrollHeight;
+      }
+
+      if (data.done) {
+        if (data.success) {
+          statusEl?.setAttribute("class", "fetch-status fetch-status-ok");
+          if (statusEl) statusEl.textContent = "\u2713 done";
+          card?.classList.add("fetch-card-ok");
+        } else {
+          statusEl?.setAttribute("class", "fetch-status fetch-status-err");
+          if (statusEl) statusEl.textContent = "\u2717 failed";
+          card?.classList.add("fetch-card-err");
+          if (data.error && logEl) {
+            logEl.innerHTML += `<div class="log-line log-line-err">${esc(data.error)}</div>`;
+            logEl.scrollTop = logEl.scrollHeight;
+          }
+        }
+      } else {
+        setTimeout(() => pollFetch(id), 2000);
+      }
+    })
+    .catch(() => setTimeout(() => pollFetch(id), 3000));
+}
+
+// ── Fetch list polling ─────────────────────────────────────────────────────────
+
+// Polls /discover/fetch/list every 5 s to discover server-created fetch entries
+// (e.g. per-album cards spawned during an artist fetch) and create cards for them.
+function initFetchList() {
+  pollFetchList();
+}
+
+function pollFetchList() {
+  fetch("/discover/fetch/list")
+    .then((r) => (r.ok ? r.json() : null))
+    .then((items) => {
+      if (!items) return;
+      for (const item of items) {
+        if (!knownFetchIds.has(item.id)) {
+          knownFetchIds.add(item.id);
+          addFetchCard(item.id, item.title);
+          if (!item.done) pollFetch(item.id);
+        }
+      }
+    })
+    .catch(() => {})
+    .finally(() => setTimeout(pollFetchList, 5000));
+}
+
+// ── Utilities ──────────────────────────────────────────────────────────────────
+
+function showFetchError(msg) {
+  const list = document.getElementById("fetch-list");
+  const el = document.createElement("div");
+  el.className = "fetch-card fetch-card-err";
+  el.innerHTML = `<div class="fetch-header">
+    <span class="fetch-title">Fetch failed</span>
+    <span class="fetch-status fetch-status-err">\u2717 error</span>
+  </div>
+  <div class="fetch-log"><div class="log-line log-line-err">${esc(msg)}</div></div>`;
+  list.prepend(el);
+}
+
+function esc(s) {
+  return String(s ?? "")
+    .replace(/&/g, "&amp;")
+    .replace(/</g, "&lt;")
+    .replace(/>/g, "&gt;")
+    .replace(/"/g, "&quot;");
+}
diff --git a/static/style.css b/static/style.css
new file mode 100644
index 0000000..4ab9e1a
--- /dev/null
+++ b/static/style.css
@@ -0,0 +1,585 @@
+/* ── Custom properties ────────────────────────────────────────────────────── */
+
+:root {
+    --bg: #111;
+    --surface: #1a1a1a;
+    --surface-hi: #222;
+    --border: #2a2a2a;
+    --border-focus: #555;
+
+    --text: #eee;
+    --text-secondary: #aaa;
+    --text-muted: #777;
+    --text-dim: #555;
+
+    --green: #4caf50;
+    --green-bg: #1e4d2b;
+    --green-hover: #1e3d1e;
+    --green-border: #3a7a3a;
+    --amber: #f0a500;
+    --amber-bg: #4d3a00;
+    --red: #e05050;
+    --red-bg: #4d1a1a;
+    --red-text: #c0392b;
+
+    --pill-beets: #7ec8e3;
+    --pill-mb: #c084fc;
+    --pill-tags: #f0a500;
+
+    --radius-lg: 8px;
+    --radius: 6px;
+    --radius-sm: 5px;
+    --radius-xs: 4px;
+
+    --max-w: 860px;
+    --pad-x: 24px;
+}
+
+/* ── Reset & base ─────────────────────────────────────────────────────────── */
+
+*,
+*::before,
+*::after {
+    box-sizing: border-box;
+}
+
+body {
+    font-family:
+        system-ui,
+        -apple-system,
+        sans-serif;
+    background: var(--bg);
+    color: var(--text);
+    margin: 0;
+    padding: 48px var(--pad-x) 80px;
+    text-align: center;
+}
+
+h1 {
+    margin: 0 0 24px;
+    font-size: clamp(20px, 4vw, 28px);
+}
+
+/* ── Tabs ─────────────────────────────────────────────────────────────────── */
+
+.tabs {
+    display: inline-flex;
+    gap: 4px;
+    background: var(--surface);
+    border: 1px solid var(--border);
+    border-radius: var(--radius-lg);
+    padding: 4px;
+    margin-bottom: 36px;
+}
+
+.tab-btn {
+    font-size: 14px;
+    min-height: 36px;
+    padding: 0 24px;
+    border-radius: var(--radius);
+    border: none;
+    cursor: pointer;
+    background: transparent;
+    color: var(--text-muted);
+    transition:
+        background 0.15s,
+        color 0.15s;
+}
+.tab-btn.active {
+    background: var(--surface-hi);
+    color: var(--text);
+}
+
+.tab-pane {
+    display: none;
+}
+.tab-pane.active {
+    display: block;
+}
+
+/* ── Shared card / content container ─────────────────────────────────────── */
+
+.content-box {
+    max-width: var(--max-w);
+    margin: 0 auto;
+    text-align: left;
+}
+
+/* ── Import tab — run button ─────────────────────────────────────────────── */
+
+.run-btn {
+    font-size: clamp(18px, 4vw, 28px);
+    padding: 18px 40px;
+    border-radius: 10px;
+    border: none;
+    cursor: pointer;
+    background: var(--green);
+    color: #fff;
+    transition: opacity 0.15s;
+}
+.run-btn:hover:not(:disabled) {
+    opacity: 0.88;
+}
+.run-btn:disabled {
+    background: #555;
+    cursor: not-allowed;
+}
+
+/* ── Import tab — session summary ────────────────────────────────────────── */
+
+.session {
+    margin-top: 48px;
+}
+
+.session-header {
+    display: flex;
+    justify-content: space-between;
+    align-items: baseline;
+    flex-wrap: wrap;
+    gap: 4px;
+    border-bottom: 1px solid #333;
+    padding-bottom: 8px;
+    margin-bottom: 20px;
+}
+.session-header h2 {
+    margin: 0;
+    font-size: 16px;
+    color: var(--text-secondary);
+}
+.session-header .duration {
+    font-size: 13px;
+    color: var(--text-dim);
+}
+
+/* ── Album card ───────────────────────────────────────────────────────────── */
+
+.album {
+    background: var(--surface);
+    border: 1px solid var(--border);
+    border-radius: var(--radius-lg);
+    padding: 16px 20px;
+    margin-bottom: 12px;
+}
+
+.album-header {
+    display: flex;
+    align-items: center;
+    gap: 10px;
+    margin-bottom: 10px;
+    flex-wrap: wrap;
+}
+.album-name {
+    font-weight: 600;
+    font-size: 15px;
+    flex: 1;
+    min-width: 0;
+    white-space: nowrap;
+    overflow: hidden;
+    text-overflow: ellipsis;
+}
+
+.badge {
+    font-size: 11px;
+    font-weight: 700;
+    padding: 2px 8px;
+    border-radius: var(--radius-xs);
+    white-space: nowrap;
+    flex-shrink: 0;
+}
+.badge-ok {
+    background: var(--green-bg);
+    color: var(--green);
+}
+.badge-warn {
+    background: var(--amber-bg);
+    color: var(--amber);
+}
+.badge-fatal {
+    background: var(--red-bg);
+    color: var(--red);
+}
+
+/* ── Metadata row ─────────────────────────────────────────────────────────── */
+
+.metadata {
+    display: flex;
+    align-items: baseline;
+    flex-wrap: wrap;
+    gap: 10px;
+    font-size: 12px;
+    color: var(--text-muted);
+    margin-bottom: 12px;
+}
+.metadata-title {
+    color: var(--text-secondary);
+    font-size: 13px;
+}
+
+.metadata-pill {
+    display: inline-flex;
+    align-items: center;
+    gap: 4px;
+    background: var(--surface-hi);
+    border-radius: var(--radius-xs);
+    padding: 2px 7px;
+    font-size: 11px;
+}
+.pill-label {
+    color: var(--text-dim);
+}
+.pill-beets {
+    color: var(--pill-beets);
+}
+.pill-musicbrainz {
+    color: var(--pill-mb);
+}
+.pill-file_tags {
+    color: var(--pill-tags);
+}
+.pill-unknown {
+    color: #888;
+}
+
+/* ── Info grid ────────────────────────────────────────────────────────────── */
+
+.info-grid {
+    display: grid;
+    grid-template-columns: repeat(auto-fill, minmax(160px, 1fr));
+    gap: 6px;
+    margin-bottom: 12px;
+}
+.info-card {
+    background: var(--surface-hi);
+    border-radius: var(--radius);
+    padding: 8px 12px;
+    font-size: 12px;
+}
+.info-card-label {
+    font-size: 10px;
+    text-transform: uppercase;
+    letter-spacing: 0.06em;
+    color: var(--text-dim);
+    margin-bottom: 4px;
+}
+.info-card-value {
+    color: var(--text-secondary);
+    font-size: 13px;
+    font-weight: 600;
+}
+.info-card-sub {
+    margin-top: 3px;
+    color: var(--text-dim);
+    font-size: 11px;
+    line-height: 1.4;
+}
+
+.info-ok {
+    color: var(--green);
+}
+.info-warn {
+    color: var(--amber);
+}
+.info-dim {
+    color: var(--text-dim);
+}
+
+/* ── Pipeline steps ───────────────────────────────────────────────────────── */
+
+.steps-label {
+    font-size: 10px;
+    text-transform: uppercase;
+    letter-spacing: 0.06em;
+    color: #444;
+    margin-bottom: 6px;
+}
+.steps {
+    display: grid;
+    grid-template-columns: repeat(auto-fill, minmax(180px, 1fr));
+    gap: 6px;
+}
+.step {
+    font-size: 12px;
+    padding: 5px 10px;
+    border-radius: var(--radius-sm);
+    background: var(--surface-hi);
+    display: flex;
+    flex-direction: column;
+    gap: 2px;
+}
+.step-label {
+    color: #888;
+}
+.step-ok {
+    color: var(--green);
+}
+.step-warn {
+    color: var(--amber);
+}
+.step-fatal {
+    color: var(--red);
+}
+.step-err {
+    font-size: 11px;
+    color: var(--red-text);
+    margin-top: 2px;
+    word-break: break-word;
+}
+
+/* ── Discover tab — search form ───────────────────────────────────────────── */
+
+.search-form {
+    display: flex;
+    gap: 8px;
+    align-items: stretch;
+    margin-bottom: 20px;
+}
+
+.type-toggle {
+    display: flex;
+    border: 1px solid #333;
+    border-radius: var(--radius);
+    overflow: hidden;
+    flex-shrink: 0;
+}
+.type-btn {
+    font-size: 13px;
+    padding: 0 16px;
+    border: none;
+    background: var(--surface);
+    color: var(--text-muted);
+    cursor: pointer;
+    transition:
+        background 0.15s,
+        color 0.15s;
+    white-space: nowrap;
+}
+.type-btn.active {
+    background: var(--surface-hi);
+    color: var(--text);
+}
+
+.search-input {
+    flex: 1;
+    min-width: 0;
+    font-size: 14px;
+    padding: 0 12px;
+    height: 38px;
+    background: var(--surface);
+    border: 1px solid #333;
+    border-radius: var(--radius);
+    color: var(--text);
+    outline: none;
+    transition: border-color 0.15s;
+}
+.search-input:focus {
+    border-color: var(--border-focus);
+}
+
+.search-btn {
+    font-size: 14px;
+    padding: 0 20px;
+    height: 38px;
+    border-radius: var(--radius);
+    border: none;
+    background: var(--green);
+    color: #fff;
+    cursor: pointer;
+    flex-shrink: 0;
+    transition: opacity 0.15s;
+}
+.search-btn:hover:not(:disabled) {
+    opacity: 0.88;
+}
+.search-btn:disabled {
+    background: #555;
+    cursor: not-allowed;
+}
+
+/* ── Discover tab — search results ───────────────────────────────────────── */
+
+.search-msg {
+    text-align: center;
+    color: var(--text-dim);
+    font-size: 14px;
+    padding: 32px 0;
+}
+.search-msg.error {
+    color: var(--red);
+}
+
+.result-row {
+    display: flex;
+    align-items: center;
+    gap: 12px;
+    background: var(--surface);
+    border: 1px solid var(--border);
+    border-radius: var(--radius-lg);
+    padding: 12px 16px;
+    margin-bottom: 8px;
+}
+.result-cover {
+    width: 48px;
+    height: 48px;
+    object-fit: cover;
+    border-radius: var(--radius-sm);
+    flex-shrink: 0;
+    background: var(--surface-hi);
+}
+.result-info {
+    flex: 1;
+    min-width: 0;
+}
+.result-title {
+    display: block;
+    font-size: 14px;
+    color: #ddd;
+    white-space: nowrap;
+    overflow: hidden;
+    text-overflow: ellipsis;
+}
+.result-dis {
+    color: var(--text-dim);
+}
+.result-meta {
+    display: block;
+    font-size: 12px;
+    color: var(--text-dim);
+    margin-top: 2px;
+}
+
+.fetch-btn {
+    font-size: 12px;
+    padding: 5px 14px;
+    border-radius: var(--radius-sm);
+    border: 1px solid var(--green-border);
+    background: transparent;
+    color: var(--green);
+    cursor: pointer;
+    flex-shrink: 0;
+    white-space: nowrap;
+    transition: background 0.15s;
+}
+.fetch-btn:hover:not(:disabled) {
+    background: var(--green-hover);
+}
+.fetch-btn:disabled {
+    border-color: #333;
+    color: var(--text-dim);
+    cursor: not-allowed;
+}
+
+/* ── Discover tab — fetch log cards ───────────────────────────────────────── */
+
+.fetch-list {
+    margin-top: 32px;
+}
+
+.fetch-card {
+    background: var(--surface);
+    border: 1px solid var(--border);
+    border-radius: var(--radius-lg);
+    padding: 14px 16px;
+    margin-bottom: 10px;
+    transition: border-color 0.3s;
+}
+.fetch-card-ok {
+    border-color: var(--green-bg);
+}
+.fetch-card-err {
+    border-color: var(--red-bg);
+}
+
+.fetch-header {
+    display: flex;
+    justify-content: space-between;
+    align-items: center;
+    flex-wrap: wrap;
+    gap: 6px;
+    margin-bottom: 8px;
+}
+.fetch-title {
+    font-size: 14px;
+    font-weight: 600;
+    color: var(--text-secondary);
+    min-width: 0;
+    overflow: hidden;
+    text-overflow: ellipsis;
+    white-space: nowrap;
+}
+.fetch-status {
+    font-size: 12px;
+    color: var(--text-dim);
+    flex-shrink: 0;
+}
+.fetch-status-ok {
+    color: var(--green);
+}
+.fetch-status-err {
+    color: var(--red);
+}
+
+.fetch-log {
+    font-size: 12px;
+    font-family: ui-monospace, "Cascadia Code", "Fira Mono", monospace;
+    color: var(--text-muted);
+    max-height: 260px;
+    overflow-y: auto;
+    scrollbar-width: thin;
+    scrollbar-color: #333 transparent;
+}
+.log-line {
+    padding: 1px 0;
+    line-height: 1.5;
+}
+.log-line-err {
+    color: var(--red-text);
+}
+
+/* ── Footer ───────────────────────────────────────────────────────────────── */
+
+footer {
+    position: fixed;
+    bottom: 14px;
+    left: 0;
+    width: 100%;
+    font-size: 12px;
+    color: #444;
+    text-align: center;
+    pointer-events: none;
+}
+
+/* ── Responsive ───────────────────────────────────────────────────────────── */
+
+@media (max-width: 600px) {
+    body {
+        padding: 32px 16px 72px;
+    }
+
+    .tabs {
+        display: flex;
+        width: 100%;
+    }
+    .tab-btn {
+        flex: 1;
+        padding: 0;
+        min-height: 40px;
+    }
+
+    .search-form {
+        flex-wrap: wrap;
+    }
+    .type-toggle {
+        width: 100%;
+    }
+    .type-btn {
+        flex: 1;
+        min-height: 38px;
+    }
+    .search-btn {
+        width: 100%;
+    }
+
+    .result-title {
+        white-space: normal;
+    }
+}