Star Trek: The Next Generation notes

• episodes.md: every episode by season with code, airdate, director, writer, arcs and a one-line synopsis.
• character_index.md: 203 characters with species, affiliation and status, linked to individual notes.
• arcs.md: episodes grouped by storyline or theme (Borg, Klingon politics, Q, Data's humanity, and so on).
• continuity.md: which episodes pick up threads from earlier ones.
• relationships.md: curated relationships plus the most frequent scene pairings.
• visuals.html: scene and episode co-appearance matrices for the main cast.
• locations.md: most used sets, by season, with a page per major location under locations/.
• entities.md: mention counts for recurring ships, places, factions and technology.
• stats.md: dialogue per character per season, catchphrase counts, longest and shortest episodes.
• actors.md: actors with multiple roles or many episodes, from Memory Alpha cast lists.
• crew.md: directors and writers, their episode counts, seasons, favourite arcs and pairings.

• episodes/<prod>.md: metadata, synopsis, continuity, every log entry, speaking characters with line counts, and scene locations.
• characters/<slug>.md: profile, key facts, rank timeline, relationships, most shared scenes, lines per season, full appearance list.

Episodes are identified on the pages by their code (S03E15, broadcast order) and overall episode number. Two-part stories with a single transcript carry a code range (S01E01-E02). The studio production number (e.g. 163) is used only as the file name for transcripts, data and notes.

• Appearance counts are based on dialogue. For the main and recurring cast, episodes where the character is only named in stage directions are listed separately as silent appearances; this is a heuristic and can miss or over-count.
• Cast lists come from Memory Alpha (CC BY-NC) and are matched to transcript labels by name, which fails for some credits.
• Speaker labels follow the transcriber's spelling. A few aliases are merged (Geordi, Beverly, Alyssa, and Data's personas in Masks).

The project was built with Claude, an AI assistant, working from my direction and review.

• Code: the scripts, the site and the build were written with Claude's help.
• Written from the transcripts: each episode's synopsis and arc tags were written by reading its full transcript.
• Written from general knowledge, then sourced: key facts, profiles, relationships and guest descriptions were first written from general knowledge of the show, not extracted from the text. Each was then checked against Memory Alpha and the episode transcripts, and corrected where they didn't match.
• Continuity links: each link between episodes was checked against both transcripts, and links with no explicit connection in the text were removed.
• Source links: a claim is followed by a link to what it rests on. That is an episode transcript (a small code such as S04E02, sometimes going to the exact line, which is highlighted when you arrive) or, for things no episode shows, such as "chief engineer from season 2", a Memory Alpha page. Airdates, directors and writers link to Wikipedia and Memory Alpha, and actors to their Memory Alpha pages. Every character fact, description and relationship has at least one source. A few clauses that neither a transcript nor Memory Alpha shows, such as Worf being "the first Klingon officer in Starfleet", have no link.
• What the build checks: that each cited episode or Memory Alpha page exists and contains the names and quoted phrases in the claim, that a quoted line is in the transcript, that a character's facts cite only episodes the character is in, and that a relationship's source names both people. This catches wrong citations. It does not prove that every sentence is true, and many of the quoted lines were chosen by Claude.
• Not checked: the choice of which entities and catchphrases to count. Character pages and portraits were audited by comparing each page's opening text with the description and by eye, so a lookalike page can still slip through. Check anything you rely on.

python3 scripts/fetch_transcripts.py   # downloads transcripts, skips files already present
python3 scripts/fetch_metadata.py      # credits and numbering from Wikipedia
python3 scripts/fetch_cast.py          # cast lists from Memory Alpha (skips pages already saved)
python3 scripts/build_notes.py         # everything under notes/

The curated content (synopses, arcs, continuity notes, profiles, relationships) lives in scripts/curated_episodes.py and scripts/curated_characters.py; edit those and rebuild.