Jess spotted that dates looked like file dates rather than library-add dates.
He is right, and MediaShelf was not the culprit: it reproduces Plex's addedAt
exactly (verified 500/500 identical to the second). Plex's own field is what
follows the file — replace or re-encode one and Date Added resets while the
item, its ratingKey and its watch history all survive.
Measured on the live library, comparing addedAt against lastViewedAt where both
exist: 55 of 509 movies (10.8%) and 306 of 1,393 TV Show Archive items (22.0%)
were watched BEFORE they were "added" — 19% overall. 2001: A Space Odyssey
reports added 2026-07-31, last watched 2017-08-26.
That is not cosmetic. pre_history is derived from added_at, so an old item whose
file was replaced looks post-coverage and gets promoted into the CONFIDENT
reclaim pool, which is the one pool meant to be trustworthy.
A completed play proves the item already existed, so added_at is now
MIN(provider_added_at, first_watched_at). Plex's raw value is kept in
provider_added_at, added_at_source records which applied, and the item drawer
explains the substitution instead of quietly disagreeing with Plex. Unwatched
items keep Plex's value since nothing contradicts it. first_seen_at is also
recorded now and is authoritative for anything added from here on.
Plex's API has no better field; the true insert time is only in Plex's own
metadata_items.created_at on Loki.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01GVbG48GAXfCZatcmX123Ra
The first real scan logged 15 "episode has no season; skipped" warnings. They
are Firefly S1 in TV Show Archive: Plex returns those episodes with
grandparentRatingKey and parentIndex set and parentGuid present, but
parentRatingKey null. Requiring parentRatingKey meant the entire season was
silently absent from the report - exactly the kind of quiet omission a reclaim
tool must not have.
The season key is now synthesized from show + season number when Plex omits it,
which is stable across scans. Keep marks are unaffected either way since they
key on GUIDs, not rating keys.
Also drops the multi_part flag from ordinary seasons. A season has one part per
episode, so part_count > 1 is normal there and the badge appeared on every TV
row; it now means what it says - a movie held more than once, or a season with
more files than episodes.
Both cases are in the fake server now, so the suite covers them.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01GVbG48GAXfCZatcmX123Ra
The application the design describes: Flask + SQLite, Plex for library data,
Tautulli for watch history, report-only.
Structure follows the design's seams. providers/ splits MediaProvider from
HistoryProvider, because on this network library data and watch data live on
different machines and Jellyfin later will have no Tautulli equivalent.
scoring.py implements the reclaim score twice - as a SQL expression for the
live grid (weights change on every slider drag, so storing it would mean
rewriting thousands of rows per drag) and in Python for CSV export and tests,
with a property test over 500 generated rows asserting the two agree.
rules.py compiles saved views to parameterized SQL through a field/operator
whitelist; nothing user-supplied is ever interpolated.
Three properties are enforced by test rather than asserted in prose:
- Ingest is idempotent. Three consecutive full scans leave every count and
every byte total unchanged. A scanner that double-counts produces a report
that looks plausible and is wrong.
- Keep marks survive Plex reassigning every rating key in the library. They
are keyed on content GUID, scoped per library so the Movies and 4K Movies
copies of the same film mark independently.
- Every config variable the app reads is declared in docker-compose.yml, so
a variable set in Portainer can never silently do nothing.
Also found and fixed while verifying against a fake Plex+Tautulli pair:
executescript() commits the pending transaction, so migrations needed their
BEGIN/COMMIT inside the script; replaceChildren() renders null as the literal
text "null"; a hash-only URL change does not reload the document, so deep
links needed a hashchange listener; and SQLite ROUND rounds half away from
zero where Python rounds half to even.
73 tests, no live server required.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01GVbG48GAXfCZatcmX123Ra
Ran the probe and a new reclaim preview against Loki and Tautulli from a LAN
host. The measured library is 65.7 TB across 24 libraries with 60 users and
87,640 logged plays. Five things in the design were wrong or missing.
- Tautulli's get_library_media_info returns file_size 0 for every show
section regardless of section_type, so it cannot cross-check TV sizes.
Plex is the only size authority for TV, which is 49 of the 66 TB.
- solitude: divisor of 3 confirmed correct (61.7% of watched items have
exactly one viewer), but weight raised 0.04 -> 0.10 since on a 60-user
server "only one person watched this" is real signal.
- rejection: abandonment is 4.9% of plays, not the 30% hoped for. Weight
cut 0.12 -> 0.06. Kept because it is decisive when it fires.
- Cross-library duplicates promoted from "later candidate" to v1: with 16
movie sections, Movies and 4K Movies routinely hold the same film.
- Protected libraries added. Family Videos is 38 GB of irreplaceable home
video that is 86% "never played" and scores as a perfect delete target.
Also splits the reclaim pool into confident (6.5 TB, added while Tautulli
was watching and never played) and uncertain (20.9 TB, predates coverage).
80% of the library predates Tautulli, so pre_history is the majority state,
not an edge case.
Adds tools/reclaim_preview.py, which produced these numbers.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01GVbG48GAXfCZatcmX123Ra
Tautulli stays local-network-only, so nothing off the LAN can check this
design against real data. tools/probe.py closes that gap from the inside:
GET requests only, standard library only, credentials redacted from all
output including error messages.
It answers open questions 1-4 in one run — Tautulli's coverage horizon,
the finished/partial/abandoned split across real plays, whether successive
plays are being grouped, library shapes and sizes, multi-version items,
and path roots by size. It also runs the pms_identifier cross-check from
section 4.11.
tools/mockserver.py mocks both APIs so the probe is testable without a live
server. Verified against it: the happy path, Tautulli absent, Tautulli
unreachable, Tautulli erroring, Plex unreachable, and an identifier
mismatch. Credential redaction confirmed in every error path.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01GVbG48GAXfCZatcmX123Ra
