fluncle 0.161.0 → 0.163.0

package/bin/fluncle.mjs

@@ -561,7 +561,7 @@ function parseVersion(version) {
 var currentVersion;
 var init_version = __esm(() => {
   init_output();
-  currentVersion = "0.161.0".trim() ? "0.161.0".trim() : "0.1.0";
+  currentVersion = "0.163.0".trim() ? "0.163.0".trim() : "0.1.0";
 });
 
 // src/update-notifier.ts
@@ -2189,7 +2189,7 @@ function parseVersion2(version) {
 var currentVersion2, latestReleaseUrl = "https://api.github.com/repos/mauricekleine/fluncle/releases/latest";
 var init_version2 = __esm(() => {
   init_output();
-  currentVersion2 = "0.161.0".trim() ? "0.161.0".trim() : "0.1.0";
+  currentVersion2 = "0.163.0".trim() ? "0.163.0".trim() : "0.1.0";
 });
 
 // ../../packages/registry/src/index.ts
@@ -2386,6 +2386,19 @@ var init_src = __esm(() => {
       url: `${SITE}/albums`,
       weights: { web: "secondary" }
     },
+    {
+      discoveryUrl: `${SITE}/llms.txt`,
+      exposedContent: [
+        "/fresh — what just came out: every drum & bass track released in the trailing 30-day window, freshest first, findings in full voice and the quieter rows in the unlit register"
+      ],
+      kind: "web_route",
+      name: "web.fresh",
+      operatorNotes: "The new-releases lens over the whole archive — the SEO answer to 'new dnb releases', a weekly-refreshed query. Orders by tracks.release_date (when a tune CAME OUT), never findings.added_at (when Fluncle FOUND it) — the two are unrelated, and the copy never claims he found these. A HUB, so it is always indexable + listed unconditionally in the sitemap (like /albums), never the per-detail thin-content gate. The window read rides the tracks_release_date_idx btree so it stays a bounded range scan as the catalogue grows (lib/server/fresh.ts). The INDEX is always-200, so it is HTTP-probeable.",
+      probeConfig: { cadenceMs: PROBE_CADENCE_MS, kind: "http", timeoutMs: PROBE_TIMEOUT_MS },
+      route: "/fresh",
+      url: `${SITE}/fresh`,
+      weights: { web: "secondary" }
+    },
     {
       exposedContent: [
         "/galaxies — the browse-by-feel lens: the archive grouped into operator-named sonic galaxies (k-means over the MuQ audio embedding space)",
@@ -2500,6 +2513,18 @@ var init_src = __esm(() => {
       url: `${SITE}/api/v1/tracks/random`,
       weights: { web: "secondary" }
     },
+    {
+      apiFormat: "application/json",
+      discoveryUrl: `${SITE}/api/v1/openapi.json`,
+      exposedContent: [
+        "what just came out — newest drum & bass releases over a 30-day window, flat (limit max 100)"
+      ],
+      kind: "api",
+      name: "api.fresh",
+      route: "/api/v1/tracks/fresh",
+      url: `${SITE}/api/v1/tracks/fresh`,
+      weights: { web: "secondary" }
+    },
     {
       apiFormat: "application/json",
       discoveryUrl: `${SITE}/api/v1/openapi.json`,
@@ -2969,6 +2994,18 @@ var init_src = __esm(() => {
       title: "Track ranking",
       weights: { status: "hidden" }
     },
+    {
+      exposedContent: [
+        "find each un-anchored catalogue row's Spotify track via Apify → its spotify_uri/spotify_url anchor"
+      ],
+      kind: "cron",
+      name: "cron.anchor",
+      operatorNotes: "hourly, run by a rave-02 HOST systemd timer (docs/agents/hermes/anchor-timer/). Fills the catalogue Spotify anchor OFF the official (dev-mode) Spotify app, which starved under 429s at catalogue scale and must stay for user-facing paths: the box runs an Apify Spotify-scraper actor to find candidates for each un-anchored catalogue row and POSTs them to the agent-tier `anchor_track` op, where the WORKER re-runs verification (exact ISRC, else the folded artist+title+±2s search triple — the box's verdict is never trusted) and writes the anchor on a hit. Every attempt stamps a 14-day re-ask backoff so a miss is not re-billed. The ONE new secret is APIFY_API_TOKEN; each row is a billed Apify search (~$0.015), so the default 15 rows/hour ≈ $5-6/day while the backlog drains — pause by stopping the timer, burn attended with `--limit N`. Calls the oRPC HTTP endpoints directly (no new CLI command the pinned box CLI would lack). Source: docs/agents/hermes/scripts/anchor-sweep.*. See docs/catalogue-crawler.md § the anchor.",
+      probeConfig: { cadenceMs: 60 * MINUTE_MS, cronName: "fluncle-anchor", kind: "cron" },
+      statusDescription: "finds each catalogue track's Spotify link",
+      title: "Spotify anchors",
+      weights: { status: "hidden" }
+    },
     {
       command: "fluncle admin frontier refresh",
       exposedContent: [
@@ -2999,6 +3036,19 @@ var init_src = __esm(() => {
       title: "Label logos",
       weights: { status: "hidden" }
     },
+    {
+      command: "fluncle admin backfills recording-mbids",
+      exposedContent: [
+        "fill each track's canonical MusicBrainz recording MBID (crawler PK strip + ISRC resolve)"
+      ],
+      kind: "cron",
+      name: "cron.recording-mbids",
+      operatorNotes: "every 60m, run by a rave-02 HOST systemd timer (docs/agents/hermes/recording-mbids-timer/). The MusicBrainz identity layer: gives every track its canonical MusicBrainz recording MBID — the one identifier that reconciles a track to the wider open music graph (MusicBrainz, Wikidata) and the anchor the `/log` MusicRecording emits as a `sameAs` + a KG `identifier`. Two fill paths: a FREE SQL strip of crawler-born rows' PK (`mb_<recording-mbid>` → the `mb_recording_id` column, no vendor call), then an ISRC→recording resolve of the findings/Spotify-born tail through the shared MusicBrainz client (`/isrc/<isrc>`). New crawler rows already carry the MBID at mint time, so this cron catches history up + drains the ISRC tail. METADATA IDENTITY ONLY — it certifies nothing and publishes nothing (agent tier, the `backfill_label_images` precedent). Worker-paced (the box holds no MusicBrainz budget): one bounded batch per tick, 1 req/s, circuit-broken on a throttle. The `tracks` row carries the durable reliability state (`mb_recording_id` + the `mb_recording_id_attempted_at` stamp, a miss stamped so it is not re-queried forever). Zero LLM tokens. Source: docs/agents/hermes/scripts/recording-mbids-sweep.*. See docs/catalogue-crawler.md.",
+      probeConfig: { cadenceMs: 60 * MINUTE_MS, cronName: "fluncle-recording-mbids", kind: "cron" },
+      statusDescription: "fills each track's MusicBrainz recording id",
+      title: "Recording MBIDs",
+      weights: { status: "hidden" }
+    },
     {
       command: "fluncle admin backfills cover-masters",
       exposedContent: [
@@ -3332,6 +3382,23 @@ var init_src = __esm(() => {
       statusDescription: "reviews + merges the nightly audit PR",
       title: "Audit reviewer",
       weights: { status: "secondary" }
+    },
+    {
+      exposedContent: [
+        "nightly Sentry triage — reads the day's unresolved production errors and opens a fix PR for each straightforward one, files the rest (claude -p, subscription auth)"
+      ],
+      kind: "cron",
+      name: "cron.sentry-triage",
+      operatorNotes: "03:30 Amsterdam, a rave-02 host systemd timer (docs/agents/hermes/sentry-triage-timer/). Its OWN cron, not the audit rotation — checks Sentry every night. A full agentic claude -p session: reconciles merged fixes (resolves their issues), pulls new unresolved issues from both Sentry projects, fixes the straightforward ones (one PR each, `Sentry-Issue:` refs so a merge resolves the issue), files the rest to docs/sentry-backlog.md. The deterministic Sentry API work lives in sentry-triage-sweep.ts so the Sentry token never enters claude. Subscription auth (CLAUDE_CODE_OAUTH_TOKEN), zero OpenRouter tokens. Operator-gated on SENTRY_TRIAGE_TOKEN (skips cleanly until set). Source: docs/agents/hermes/scripts/sentry-triage-sweep.sh. Probed on /status as service `cron.sentry-triage`.",
+      probeConfig: {
+        cadenceMs: 24 * 60 * MINUTE_MS,
+        cronName: "fluncle-sentry-triage",
+        kind: "cron",
+        schedule: { time: "03:30", tz: "Europe/Amsterdam" }
+      },
+      statusDescription: "nightly triage of production errors → fix PRs",
+      title: "Sentry triage",
+      weights: { status: "secondary" }
     }
   ];
 });
@@ -3837,6 +3904,7 @@ __export(exports_admin_tracks, {
   embedQueueCommand: () => embedQueueCommand,
   contextQueueCommand: () => contextQueueCommand,
   captureQueueCommand: () => captureQueueCommand,
+  backfillRecordingMbidsCommand: () => backfillRecordingMbidsCommand,
   backfillLastfmCommand: () => backfillLastfmCommand,
   backfillDiscogsCommand: () => backfillDiscogsCommand,
   backfillAppleMusicCommand: () => backfillAppleMusicCommand,
@@ -4024,6 +4092,13 @@ async function backfillAppleCatalogueCommand(limit, dryRun) {
   const params = new URLSearchParams({ dryRun: String(dryRun), limit: String(limit) });
   return adminApiPost(`/api/admin/backfill/apple-catalogue?${params.toString()}`);
 }
+async function backfillRecordingMbidsCommand(limit, dryRun, cursor) {
+  const params = new URLSearchParams({ dryRun: String(dryRun), limit: String(limit) });
+  if (cursor) {
+    params.set("cursor", cursor);
+  }
+  return adminApiPost(`/api/admin/backfill/recording-mbids?${params.toString()}`);
+}
 async function vehiclesCommand(limit) {
   const tracks = await fetchAdminTracks({ hasVideo: true, max: limit, order: "desc" });
   return tracks.map((track) => ({
@@ -9358,6 +9433,10 @@ function addAdminCommands(program2) {
     const { backfillLabelImagesCommand: backfillLabelImagesCommand2 } = await Promise.resolve().then(() => (init_admin_labels(), exports_admin_labels));
     await runBackfillLabelImages(options, backfillLabelImagesCommand2);
   });
+  backfill.command("recording-mbids").description("Fill MusicBrainz recording MBIDs (crawler PK strip + ISRC resolve) over tracks").option("--dry-run", "Report the eligible worklist without any vendor call or write", false).option("--limit <limit>", "Max ISRC lookups to process", "50").option("--json", "Print JSON", false).action(async (options) => {
+    const { backfillRecordingMbidsCommand: backfillRecordingMbidsCommand2 } = await Promise.resolve().then(() => (init_admin_tracks(), exports_admin_tracks));
+    await runBackfillRecordingMbids(options, backfillRecordingMbidsCommand2);
+  });
   backfill.command("cover-masters").description("Resolve owned \u22641200\xB2 cover masters (album/artist) into R2").option("--kind <kind>", "Which entity to drain: album | artist", "album").option("--dry-run", "Report the eligible worklist without any fetch or write", false).option("--limit <limit>", "Max entities to process", "50").option("--json", "Print JSON", false).action(async (options) => {
     const { backfillCoverMastersCommand: backfillCoverMastersCommand2 } = await Promise.resolve().then(() => (init_admin_covers(), exports_admin_covers));
     await runBackfillCoverMasters(options, backfillCoverMastersCommand2);
@@ -9779,18 +9858,6 @@ async function runBackfillLastfm(options, backfillLastfmCommand2) {
     process.exitCode = 1;
   }
 }
-function anchorOutcomeSuffix(outcome) {
-  switch (outcome) {
-    case "throttled":
-      return " (Spotify throttled the fill; the anchor breaker is pacing it)";
-    case "unauthorized":
-      return " (Spotify grant is gone \u2014 reconnect Spotify from /admin to resume anchoring)";
-    case "breaker_open":
-      return " (anchor fill PAUSED on its breaker; it resumes after the cooldown)";
-    default:
-      return "";
-  }
-}
 async function runCrawlCatalogue(options, crawlCatalogueCommand2) {
   const limit = parseListLimit(options.limit);
   const maxHop = Number.parseInt(options.maxHop ?? "2", 10);
@@ -9802,7 +9869,6 @@ async function runCrawlCatalogue(options, crawlCatalogueCommand2) {
   } else {
     console.log(`Expanded ${result.expanded} node(s) at hop \u2264 ${result.maxHop}: ${result.tracksFound} track(s) found, ${result.tracksWritten} written, ${result.tracksSkipped} already held.`);
     console.log(`  +${result.seeded} seed(s), +${result.nodesEnqueued} node(s) enqueued, ${result.frontierPending} pending, ${result.failed} failed.`);
-    console.log(`  ${result.anchorsFilled} Spotify anchor(s) filled${anchorOutcomeSuffix(result.anchorOutcome)}.`);
     if (result.seedsRearmed > 0) {
       console.log(`  Re-armed ${result.seedsRearmed} enabled seed label(s) to re-check for new releases.`);
     }
@@ -9825,14 +9891,6 @@ async function runCrawlStatus(options, crawlStatusCommand2) {
   }
   const { frontier, frontierByKind } = result;
   console.log(`Catalogue: ${result.catalogueTracks} uncertified track(s); ${result.anchorsPending} awaiting a Spotify anchor.`);
-  const { spotifyAnchor } = result;
-  if (spotifyAnchor.tripped) {
-    const mins = Math.ceil(spotifyAnchor.cooldownRemainingMs / 60000);
-    const why = spotifyAnchor.reason === "unauthorized" ? "the Spotify grant is gone \u2014 reconnect Spotify from /admin to resume anchoring" : "Spotify is throttling the app \u2014 it self-heals when the throttle lifts";
-    console.log(`  Spotify anchor PAUSED (${why}); breaker cools in ~${mins} min.`);
-  } else if (spotifyAnchor.consecutiveFailures > 0) {
-    console.log(`  Spotify anchor wobbling (${spotifyAnchor.consecutiveFailures} failing pass(es); reason: ${spotifyAnchor.reason ?? "unknown"}).`);
-  }
   console.log(`Frontier: ${frontier.pending} pending, ${frontier.done} done, ${frontier.failed} failed, ${frontier.skipped} skipped (${frontierByKind.label} label, ${frontierByKind.artist} artist, ${frontierByKind.release} release).`);
   console.log(`Seeds (${result.seedLabels.length}): ${result.seedLabels.join(", ") || "none enabled"}.`);
   console.log(`Awaiting your ruling: ${result.labelsUndecided} label(s).`);
@@ -10230,6 +10288,63 @@ async function runBackfillLabelImages(options, backfillLabelImagesCommand2) {
     process.exitCode = 1;
   }
 }
+async function runBackfillRecordingMbids(options, backfillRecordingMbidsCommand2) {
+  const limit = parseListLimit(options.limit);
+  const resolved = [];
+  const missed = [];
+  const failed = [];
+  let cursor;
+  let dryRun = options.dryRun;
+  let prefixStripped = 0;
+  let throttled = false;
+  while (resolved.length + missed.length + failed.length < limit) {
+    const remaining = limit - (resolved.length + missed.length + failed.length);
+    const result = await backfillRecordingMbidsCommand2(remaining, options.dryRun, cursor);
+    dryRun = result.dryRun;
+    prefixStripped += result.prefixStripped;
+    resolved.push(...result.resolved);
+    missed.push(...result.missed);
+    failed.push(...result.failed);
+    if (!options.json) {
+      const verb2 = result.dryRun ? "would resolve" : "resolved";
+      console.log(`  \u2026stripped ${result.prefixStripped} from PK; ${verb2} ${result.resolvedCount} by ISRC; ${result.missedCount} not in MusicBrainz; ${result.failedCount} failed`);
+    }
+    if (result.rateLimited) {
+      throttled = true;
+      break;
+    }
+    if (result.nextCursor === null) {
+      break;
+    }
+    cursor = result.nextCursor;
+  }
+  if (options.json) {
+    printJson({
+      dryRun,
+      failed,
+      failedCount: failed.length,
+      missed,
+      missedCount: missed.length,
+      ok: true,
+      prefixStripped,
+      rateLimited: throttled,
+      resolved,
+      resolvedCount: resolved.length
+    });
+    if (failed.length > 0) {
+      process.exitCode = 1;
+    }
+    return;
+  }
+  const verb = dryRun ? "Would fill" : "Filled";
+  console.log(`${verb} ${prefixStripped} recording MBID(s) from crawler PKs; ${resolved.length} by ISRC; ${missed.length} not in MusicBrainz; ${failed.length} failed.`);
+  for (const item of failed) {
+    console.log(`  ${item.trackId}: ${item.error}`);
+  }
+  if (failed.length > 0) {
+    process.exitCode = 1;
+  }
+}
 async function runArtistResolveQueue(options, listArtistsCommand2) {
   const limit = parseListLimit(options.limit);
   const result = await listArtistsCommand2(limit ?? 50);

package/package.json

@@ -31,5 +31,5 @@
     "url": "git+https://github.com/mauricekleine/fluncle.git"
   },
   "type": "module",
-  "version": "0.161.0"
+  "version": "0.163.0"
 }