personal memory agent
Auto-skip trashed and short recordings during Plaud sync
Use is_trash and duration fields from the Plaud API to automatically
mark junk files as skipped instead of available. Trashed recordings
and those under 30 seconds are filtered out with auditable skip_reason.
- Store richer metadata (duration, is_trash) in sync state entries
- Show skipped count in CLI summary output
- Update migration script with same filtering logic
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
+81
-3
+59
tests/test_importer_sync.py
+59
tests/test_importer_sync.py
···
238
238
239
239
assert result["total"] == 2
240
240
assert result["available"] == 2
241
+
assert result["skipped"] == 0
241
242
assert result["imported"] == 0
242
243
assert result["downloaded"] == 0
243
244
···
351
352
state = load_sync_state(tmp_path, "plaud")
352
353
assert state["files"]["file2"]["status"] == "imported"
353
354
assert state["files"]["file2"]["import_timestamp"] == "20260117_134640"
355
+
356
+
357
+
def _mock_list_files_with_junk(_session, _token):
358
+
"""Return a file list including trashed and short recordings."""
359
+
return [
360
+
{
361
+
"id": "good1",
362
+
"filename": "Team Standup",
363
+
"fullname": "aaa.opus",
364
+
"filesize": 5000,
365
+
"start_time": 1737000000000,
366
+
"duration": 300000,
367
+
"is_trash": False,
368
+
},
369
+
{
370
+
"id": "trashed1",
371
+
"filename": "Old Recording",
372
+
"fullname": "bbb.opus",
373
+
"filesize": 2000,
374
+
"start_time": 1737100000000,
375
+
"duration": 60000,
376
+
"is_trash": True,
377
+
},
378
+
{
379
+
"id": "short1",
380
+
"filename": "Accidental Tap",
381
+
"fullname": "ccc.opus",
382
+
"filesize": 500,
383
+
"start_time": 1737200000000,
384
+
"duration": 5000,
385
+
"is_trash": False,
386
+
},
387
+
]
388
+
389
+
390
+
def test_plaud_sync_skips_trashed_and_short(tmp_path, monkeypatch):
391
+
"""Trashed and short recordings are auto-skipped."""
392
+
from think.importers.plaud import PlaudBackend
393
+
from think.importers.sync import load_sync_state
394
+
395
+
monkeypatch.setenv("PLAUD_ACCESS_TOKEN", "test-token")
396
+
397
+
with patch(
398
+
"think.importers.plaud.list_files", side_effect=_mock_list_files_with_junk
399
+
):
400
+
result = PlaudBackend().sync(tmp_path, dry_run=True)
401
+
402
+
assert result["total"] == 3
403
+
assert result["available"] == 1
404
+
assert result["skipped"] == 2
405
+
assert result["imported"] == 0
406
+
407
+
state = load_sync_state(tmp_path, "plaud")
408
+
assert state["files"]["good1"]["status"] == "available"
409
+
assert state["files"]["trashed1"]["status"] == "skipped"
410
+
assert state["files"]["trashed1"]["skip_reason"] == "trashed"
411
+
assert state["files"]["short1"]["status"] == "skipped"
412
+
assert state["files"]["short1"]["skip_reason"] == "too_short"
354
413
355
414
356
415
def test_plaud_sync_cli_flag(capsys, monkeypatch, tmp_path):
+5
-2
think/importers/cli.py
+5
-2
think/importers/cli.py
···
124
124
total = result.get("total", 0)
125
125
imported = result.get("imported", 0)
126
126
available = result.get("available", 0)
127
+
skipped = result.get("skipped", 0)
127
128
downloaded = result.get("downloaded", 0)
128
129
errors = result.get("errors", [])
129
130
130
131
# Print summary
131
132
print()
132
-
print(f" Total recordings: {total}")
133
-
print(f" Already imported: {imported}")
133
+
print(f" Total recordings: {total}")
134
+
print(f" Already imported: {imported}")
134
135
print(f" Available to import: {available}")
136
+
if skipped:
137
+
print(f" Skipped: {skipped} (trashed/short)")
135
138
136
139
if downloaded > 0:
137
140
print(f" Downloaded + imported: {downloaded}")
+17
-1
think/importers/plaud.py
+17
-1
think/importers/plaud.py
···
23
23
24
24
API_BASE = "https://api.plaud.ai"
25
25
26
+
# Skip recordings shorter than this (milliseconds)
27
+
MIN_DURATION_MS = 30_000
28
+
26
29
27
30
def make_session() -> requests.Session:
28
31
"""Create a requests session with sane retries."""
···
316
319
entry["matched_at"] = dt.datetime.now().isoformat()
317
320
continue
318
321
319
-
# New file — check if matched to an existing import
322
+
# New file — build entry with full metadata
323
+
duration = file_info.get("duration", 0)
324
+
is_trash = file_info.get("is_trash", False)
325
+
320
326
entry: dict[str, Any] = {
321
327
"filename": file_info.get("filename", "unnamed"),
322
328
"fullname": file_info.get("fullname", ""),
323
329
"filesize": file_info.get("filesize", 0),
324
330
"start_time": file_info.get("start_time", 0),
331
+
"duration": duration,
332
+
"is_trash": is_trash,
325
333
}
326
334
327
335
if file_id in matches:
328
336
entry["status"] = "imported"
329
337
entry["import_timestamp"] = matches[file_id]
330
338
entry["matched_at"] = dt.datetime.now().isoformat()
339
+
elif is_trash:
340
+
entry["status"] = "skipped"
341
+
entry["skip_reason"] = "trashed"
342
+
elif duration and duration < MIN_DURATION_MS:
343
+
entry["status"] = "skipped"
344
+
entry["skip_reason"] = "too_short"
331
345
else:
332
346
entry["status"] = "available"
333
347
···
339
353
available = sum(
340
354
1 for f in known_files.values() if f.get("status") == "available"
341
355
)
356
+
skipped = sum(1 for f in known_files.values() if f.get("status") == "skipped")
342
357
343
358
result: dict[str, Any] = {
344
359
"total": total,
345
360
"imported": imported,
346
361
"available": available,
362
+
"skipped": skipped,
347
363
"downloaded": 0,
348
364
"errors": [],
349
365
}