Watch history

Exports

Streamline reads the data exports each service gives you under privacy law. None of these services has a public API for watch history, and scraping breaks their terms and breaks often. The exports are legitimate, stable, and complete.

Service	Path in config.local.yaml	How to get it
Netflix	platform_paths.netflix	Account Settings > Download your data
Prime Video	platform_paths.prime	Account > Digital content > Request your data
Apple TV	platform_paths.apple_tv	Apple’s privacy site: Apple Media Services Information
Disney+	platform_paths.disney	A personal data request to Disney; the history arrives as a PDF
Max	platform_paths.hbo	A personal data request to Warner Bros. Discovery; the history arrives as a folder of CSV files

Point each path at the export exactly as downloaded: the zip for Netflix, Prime Video, and Apple TV, the PDF for Disney+, and the unzipped folder for Max. Each path can also be a list, for more than one export from the same service.

A few differences to know:

• Disney+ exports every profile on the account. Set disney_profiles in config.local.yaml to the profile names to keep, for example disney_profiles: ["Alex"].
• Max exports have no watch dates, so Max titles sort last in Archive’s Recently watched, and each title counts once. After adding a newer export, run ./recommend setup --refresh-data.

A routine data refresh never rebuilds the taste profile. It does write descriptions for new titles, which are fast-model calls.

Plex

Plex sends plays as they happen; see Plex.

Manual lists

Put one title per line in data/manual/tv.csv and data/manual/movies.csv. A movie title may end with its year (Zodiac 2007), which helps matching. You can also add a single title with ./recommend --add "Shetland" --type tv, or Add watched in the Archive.

Fixing titles

When TMDB can’t match a title, or matches the wrong one, add it to data/overrides.json:

{
  "The Matrix III Revolutions": {"title": "The Matrix Revolutions"},
  "Some Cooking Show Ep 3": {"skip": true},
  "Delhi Cops Episode": {"title": "Delhi Cops", "content_type": "tv"},
  "Specific Movie": {"tmdb_id": 12345}
}

Each entry can set the title to search for, the content_type (tv or movie), an exact tmdb_id, or skip to leave the title out. The next ./recommend setup notices the change.

Where it is stored

Watch events, ratings, the watchlist, follows, and search history all live in one SQLite file, data/streamline.db. Plex plays exist only there, so back that file up.