refactor(clips): render clean video filmstrip

scripts/i18n-raw-literal-baseline.txt

@@ -6,7 +6,6 @@ packages/docs/app/components/blocks/comparison.tsx|A side-by-side comparison of
 packages/docs/app/components/blocks/steps.tsx|A numbered step-by-step procedure with circled step numbers.
 packages/docs/app/components/blocks/steps.tsx|Step title
 packages/docs/app/components/docBlocks.tsx|| Extract
-templates/clips/app/components/editor/waveform.tsx|VISUAL_SILENCE_FLOOR && botY - topY
 templates/clips/app/components/recorder/storage-setup-card.tsx|AWS S3, Cloudflare R2, DigitalOcean Spaces, MinIO
 templates/clips/app/components/workspace/notifications-list.tsx|["formatDate"], formatRelativeTime: ReturnType
 templates/crm/app/components/crm/CreateCrmRecordDialog.tsx|This record lives in the local-authoritative Native SQL CRM. You can connect or mirror another CRM separately.

templates/clips/.agents/skills/video-editing/SKILL.md

@@ -229,6 +229,37 @@ downsamples to a 2000-point peaks array. Peaks are cached in
 recompute. The editor reads the cache first and falls back to computing only
 when the key is missing or corrupted.
 
+### Timeline filmstrip
+
+The trim track shows video frames behind the waveform. Two paths, and the order
+matters:
+
+1. **Sprite (preferred).** `generate-filmstrip` runs one ffmpeg pass
+   (`fps` → `scale` → `pad` → `tile`) into a single JPEG grid, uploads it, and
+   stores the URL plus grid geometry on the recording. The editor renders cells
+   with CSS `background-position` — one cached image request, no video decoding
+   in the browser.
+2. **Browser extraction (fallback).** `extractFilmstripThumbnails()` seeks a
+   detached `<video>` once per frame. Only used when there is no sprite, which
+   means hosts without ffmpeg and local/dev media the server cannot fetch.
+
+Three traps this design exists to avoid:
+
+- **Pass the proxied URL.** Frame extraction must use `getWaveformMediaUrl()`,
+  never `recording.videoUrl` — reading pixels back out of a cross-origin video
+  taints the canvas and `toDataURL` throws, so provider media silently yields no
+  filmstrip at all.
+- **Both paths sample cell midpoints,** not `0 … duration` endpoints. The strip
+  renders N equal cells, so cell `i` must show the middle of the slot it
+  occupies or every thumbnail sits up to half a cell from the time beneath it.
+- **Cell count comes from the geometry,** `trackWidth / (height × aspect)`, not
+  from the sprite's frame count. Rendering all 40 frames across an unzoomed
+  track makes each cell portrait, and `object-cover` then shows a narrow centre
+  slice of each frame instead of a recognisable thumbnail.
+
+The filmstrip never replaces the waveform — peaks still draw on top of a scrim,
+because frames say nothing about the audio.
+
 ### Keyboard shortcuts (editor scope)
 
 - **Space** — play / pause (even while focused in the editor area)

templates/clips/AGENTS.md

@@ -92,6 +92,7 @@ Read the matching skill before deeper work in that area:
 | `move-recording` | Move `id` or `ids` to a folder or root |
 | `archive-`, `trash-`, `restore-recording` | Lifecycle |
 | `reprocess-recording` | Repair unseekable/frozen media |
+| `generate-filmstrip` | Editor timeline frame sprite; `all` backfills |
 | `request-transcript`, `cleanup-transcript` | Transcribe; `force`/`regenerate` |
 | `regenerate-title`, `-summary`, `-chapters` | AI metadata |
 | `trim-`, `split-recording`, `remove-silences`, `remove-filler-words` | Edits |

templates/clips/actions/generate-filmstrip.ts

@@ -0,0 +1,154 @@
+/**
+ * Generate editor-timeline filmstrip sprites for stored recordings.
+ *
+ * The editor shows video frames behind the waveform so a trim range can be
+ * found visually. Generating them server-side with one ffmpeg pass replaces
+ * per-frame decoding in the browser, which cannot read cross-origin media at
+ * all and needs one seek per frame.
+ *
+ * Non-destructive and idempotent: a recording whose sprite already matches its
+ * current media is skipped, and any failure leaves the row untouched so the
+ * editor keeps its browser-side fallback.
+ *
+ * Usage:
+ *   pnpm action generate-filmstrip --id=<recordingId>
+ *   pnpm action generate-filmstrip --id=<recordingId> --force
+ *   pnpm action generate-filmstrip --all --limit=20
+ */
+
+import { defineAction } from "@agent-native/core";
+import { and, desc, eq, isNotNull, isNull } from "drizzle-orm";
+import { z } from "zod";
+
+import { getDb, schema } from "../server/db/index.js";
+import {
+  getCurrentOwnerEmail,
+  ownerEmailMatches,
+} from "../server/lib/recordings.js";
+import {
+  DEFAULT_FILMSTRIP_FRAME_COUNT,
+  MAX_FILMSTRIP_FRAME_COUNT,
+} from "../server/lib/video-filmstrip-sprite.js";
+import {
+  ensureRecordingFilmstrip,
+  type EnsureFilmstripResult,
+} from "./lib/ensure-recording-filmstrip.js";
+
+const MAX_TARGETS_PER_CALL = 50;
+const DEFAULT_ALL_LIMIT = 10;
+const cliBoolean = z.preprocess((value) => {
+  if (value === "true" || value === "1") return true;
+  if (value === "false" || value === "0") return false;
+  return value;
+}, z.boolean());
+
+export default defineAction({
+  description:
+    "Generate the editor timeline filmstrip (a sprite of evenly-spaced video frames) for one or more of the caller's recordings, so the editor can show frames behind the waveform without decoding video in the browser. Pass `id` for one clip, `ids` for several, or `all: true` to backfill clips that have no filmstrip yet. Recordings whose sprite already matches their current media are skipped unless `force` is true. Requires ffmpeg on the server; without it this reports `skipped-no-ffmpeg` and the editor falls back to browser-side extraction.",
+  schema: z.object({
+    id: z
+      .string()
+      .optional()
+      .describe("A single recording id to generate a filmstrip for."),
+    ids: z
+      .array(z.string())
+      .optional()
+      .describe("Several recording ids to process in one call."),
+    all: cliBoolean
+      .optional()
+      .describe(
+        "Backfill the caller's ready clips that have no filmstrip yet, most recent first, up to `limit`.",
+      ),
+    limit: z.coerce
+      .number()
+      .int()
+      .min(1)
+      .max(MAX_TARGETS_PER_CALL)
+      .optional()
+      .describe(
+        `Max clips to process when using \`all\` (default ${DEFAULT_ALL_LIMIT}).`,
+      ),
+    frameCount: z.coerce
+      .number()
+      .int()
+      .min(4)
+      .max(MAX_FILMSTRIP_FRAME_COUNT)
+      .optional()
+      .describe(
+        `How many frames to sample across the clip (default ${DEFAULT_FILMSTRIP_FRAME_COUNT}). More frames means finer scrubbing detail and a larger sprite.`,
+      ),
+    force: cliBoolean
+      .optional()
+      .describe("Regenerate even when a current filmstrip already exists."),
+  }),
+  run: async (args) => {
+    const db = getDb();
+    const ownerEmail = getCurrentOwnerEmail();
+
+    let targetIds: string[] = [];
+    if (args.id) targetIds.push(args.id);
+    if (args.ids?.length) targetIds.push(...args.ids.filter(Boolean));
+
+    if (args.all) {
+      const conditions = [
+        ownerEmailMatches(schema.recordings.ownerEmail, ownerEmail),
+        eq(schema.recordings.status, "ready"),
+        isNotNull(schema.recordings.videoUrl),
+      ];
+      if (!args.force) {
+        conditions.push(isNull(schema.recordings.filmstripUrl));
+      }
+      const rows = await db
+        .select({ id: schema.recordings.id })
+        .from(schema.recordings)
+        .where(and(...conditions))
+        .orderBy(desc(schema.recordings.createdAt))
+        .limit(args.limit ?? DEFAULT_ALL_LIMIT);
+      targetIds.push(...rows.map((r) => r.id));
+    }
+
+    // De-dupe while preserving order, and bound the batch so one call can't run
+    // unboundedly under the hosted foreground budget.
+    targetIds = Array.from(new Set(targetIds)).slice(0, MAX_TARGETS_PER_CALL);
+
+    if (targetIds.length === 0) {
+      return {
+        ok: true,
+        processed: 0,
+        changed: 0,
+        results: [] as EnsureFilmstripResult[],
+        message:
+          "No recordings to process. Pass id / ids, or all: true to backfill clips without a filmstrip.",
+      };
+    }
+
+    const results: EnsureFilmstripResult[] = [];
+    for (const recordingId of targetIds) {
+      try {
+        results.push(
+          await ensureRecordingFilmstrip({
+            recordingId,
+            ownerEmail,
+            frameCount: args.frameCount,
+            force: Boolean(args.force),
+          }),
+        );
+      } catch (err) {
+        console.warn("[generate-filmstrip] failed for", recordingId, err);
+        results.push({
+          recordingId,
+          status: "failed-ffmpeg",
+          changed: false,
+          detail: err instanceof Error ? err.message : String(err),
+        });
+      }
+    }
+
+    const changed = results.filter((r) => r.changed).length;
+    console.log(
+      `Processed ${results.length} recording(s); ${changed} filmstrip(s) generated.`,
+    );
+
+    return { ok: true, processed: results.length, changed, results };
+  },
+});

templates/clips/actions/get-recording-player-data.ts

@@ -314,6 +314,12 @@ export default defineAction({
         description: rec.description,
         thumbnailUrl: rec.thumbnailUrl,
         animatedThumbnailUrl: rec.animatedThumbnailUrl,
+        filmstripUrl: rec.filmstripUrl ?? null,
+        filmstripFrameCount: rec.filmstripFrameCount ?? 0,
+        filmstripColumns: rec.filmstripColumns ?? 0,
+        filmstripRows: rec.filmstripRows ?? 0,
+        filmstripFrameWidth: rec.filmstripFrameWidth ?? 0,
+        filmstripFrameHeight: rec.filmstripFrameHeight ?? 0,
         sourceAppName: rec.sourceAppName,
         sourceWindowTitle: rec.sourceWindowTitle,
         durationMs: rec.durationMs,