The Codex · Data

The database, as tables

Everything in the Codex as documented tables with keys: characters, teams, powers, creators, performers, productions and every relationship between them — 194,151 rows across 20 tables. Built for language models, researchers and developers. No key, no rate limit. CC BY 4.0 — cite Comic Book Co.

datapackage.json

every table, column, key and licence (Frictionless standard)

codex.sqlite.gz

all tables in one SQLite file · 20.2 MB gzipped

Explorer

walk the web of relationships

llms.txt

how an AI agent should use this

How to read a row

curated = true records are hand-built by Comic Book Co and pass an evidence gate; every curated ability is backed by cited feats.
curated = false records come from the data pipeline — licensed, structured sources ingested by deterministic code, never generated by AI.
Where both exist for the same fact, the evidence column says which you are reading. Do not treat a source-recorded ability as equal to a curated one.
Every row traces to statements, which carries source, tier, confidence and validity dates.

Entities 36,920 rows

Every record in the Codex, of every type. The spine every other table joins to: characters, teams, powers, people, films, shows, episodes, games, comic and manga series, publishers, studios and universes.

entities.csv · About this table

Columns

Column	Type	Meaning
`id`	string	Permanent id, '<type>:<slug>'
`type`	string	Record type
`name`	string	Display name
`level`	string	entity = the canonical record; version = that record in one continuity (Spider-Man in the MCU); variant = an in-fiction alternate
`canonical_id`	string	For versions and variants, the canonical record they belong to → `entities`
`description`	string	Short description (our summary where curated, else the source's)
`first_year`	integer	Year of first publication or release, where known
`curated`	boolean	true = hand-built and evidence-gated by Comic Book Co; false = imported by the data pipeline
`wikidata`	string	Wikidata QID

Characters 19,802 rows

Fictional characters across comics, manga, anime, film and TV, one row per canonical character.

characters.csv · About this table

Columns

Column	Type	Meaning
`id`	string	Character id → `entities`
`name`	string	Display name
`real_name`	string	Civilian or given name, where recorded
`universe`	string	Narrative universe (Earth-616, DC Universe, ...)
`gender`	string	As recorded by the source
`description`	string	Short description
`first_appearance_id`	string	Work of first appearance → `entities`
`curated`	boolean	Hand-built and evidence-gated
`wikidata`	string	Wikidata QID
`comicvine`	string	Comic Vine id

Character abilities 6,577 rows

Which characters have which powers. `evidence` separates the two kinds of row and they must not be read as equal: 'curated' rows passed the Codex evidence gate (see feats); 'source' rows are recorded by Wikidata.

character_abilities.csv · About this table

Columns

Column	Type	Meaning
`character_id`	string	Character → `entities`
`power_id`	string	Power or ability → `entities`
`evidence`	string	curated \| source

Feats 1,661 rows

The feat ledger: a character demonstrating a power in a cited work. The atomic evidence record behind every curated ability.

feats.csv · About this table

Columns

Column	Type	Meaning
`id`	integer	Statement id
`character_id`	string	Character → `entities`
`power_id`	string	Power demonstrated → `entities`
`claim`	string	What happened
`work_id`	string	Cited work → `entities`
`locator`	string	Issue, chapter or episode
`continuity_id`	string	Continuity the claim belongs to → `entities`
`year`	string	Year of the cited work
`tier`	string	depicted (shown) \| stated (said) \| databook \| authorial \| calc
`evidence`	string	What the claim rests on
`confidence`	string	high \| medium \| low — how sure we are

Character creators 8,554 rows

Who created each character.

character_creators.csv · About this table

Columns

Column	Type	Meaning
`character_id`	string	Character → `entities`
`person_id`	string	Creator → `entities`

Team memberships 4,918 rows

Characters and the teams and organisations they belong or belonged to.

team_memberships.csv · About this table

Columns

Column	Type	Meaning
`character_id`	string	Member → `entities`
`team_id`	string	Team or organisation → `entities`

Enemies 3,416 rows

Recorded antagonist relationships, as stated by the source (directional).

enemies.csv · About this table

Columns

Column	Type	Meaning
`character_id`	string	Character → `entities`
`enemy_id`	string	Their enemy → `entities`

Relatives 7,413 rows

Family relationships with the kinship named.

relatives.csv · About this table

Columns

Column	Type	Meaning
`character_id`	string	Character → `entities`
`relative_id`	string	Relative → `entities`
`kinship`	string	father, mother, sibling, spouse, child, partner, ...

First appearances 1,865 rows

The work each character first appeared in. `evidence` = curated (Codex key-issue record, with issue number) or source (Wikidata).

first_appearances.csv · About this table

Columns

Column	Type	Meaning
`character_id`	string	Character → `entities`
`work_id`	string	Work of first appearance → `entities`
`issue`	string	Issue number where recorded
`evidence`	string	curated \| source

Appearances 34,111 rows

Works a character or entity is present in (films, series, games, comics).

appearances.csv · About this table

Columns

Column	Type	Meaning
`entity_id`	string	Character or other entity → `entities`
`work_id`	string	Work → `entities`

Screen credits 18,061 rows

Who played whom, in what. Live-action and voice. `classification` marks parody and fan productions so they are never read as canon.

screen_credits.csv · About this table

Columns

Column	Type	Meaning
`character_id`	string	Character played → `entities`
`person_id`	string	Performer → `entities`
`production_id`	string	Film, show, episode or game → `entities`
`kind`	string	live \| voice
`classification`	string	canonical \| parody \| fan
`year`	string	Release year
`evidence`	string	curated \| source

Adaptations 1,371 rows

Screen and other works based on a character or story.

adaptations.csv · About this table

Columns

Column	Type	Meaning
`work_id`	string	The adaptation → `entities`
`source_id`	string	What it is based on (character or arc) → `entities`
`evidence`	string	curated \| source

Key issues 213 rows

Collector-relevant milestones: first appearances, deaths, returns, team changes.

key_issues.csv · About this table

Columns

Column	Type	Meaning
`id`	integer	Statement id
`character_id`	string	Character → `entities`
`kind`	string	first_appearance \| death \| return \| joins_team \| ...
`work_id`	string	Series → `entities`
`issue`	string	Issue number
`year`	string	Year
`note`	string	Why it matters

Corrections 191 rows

The constraint registry: widely repeated claims that are WRONG, each paired with the accurate statement. Check here before asserting a capability.

corrections.csv · About this table

Columns

Column	Type	Meaning
`subject_id`	string	What the correction is about → `entities`
`locator`	string	Pair key
`common_error`	string	The claim people and models get wrong
`correction`	string	What is actually true
`work_id`	string	Supporting work → `entities`

Universes and continuities 356 rows

Narrative universes and publishing continuities, and how many records sit in each.

universes.csv · About this table

Columns

Column	Type	Meaning
`id`	string	Universe or continuity → `entities`
`name`	string	Name
`records`	integer	Entities placed in it

External identifiers 48,722 rows

The crosswalk: the same record's id in Wikidata, Comic Vine, IMDb, TMDB, GCD, AniList and others. Join your own data to ours through these.

external_ids.csv · About this table

Columns

Column	Type	Meaning
`entity_id`	string	Codex record → `entities`
`namespace`	string	wikidata \| comicvine \| imdb \| tmdb-movie \| gcd-series \| ...
`value`	string	Identifier in that namespace

Aliases 68,356 rows

Alternative names, codenames and real names that resolve to one record.

aliases.csv · About this table

Columns

Column	Type	Meaning
`entity_id`	string	Codex record → `entities`
`name`	string	Alias
`kind`	string	alias \| real_name \| aka

All statements 206,816 rows

Every verified statement in the graph, in its native form. Everything above is a view of this table; use it for anything the views do not cover.

statements.csv · About this table

Columns

Column	Type	Meaning
`id`	integer	Statement id
`subject_id`	string	Subject → `entities`
`predicate`	string	Relation or attribute (see predicates) → `predicates`
`value`	string	Literal value
`unit`	string	Unit or qualifier
`object_id`	string	Object, when the statement is an edge → `entities`
`context_id`	string	Context (e.g. the production of a screen credit) → `entities`
`continuity_id`	string	Continuity scope → `entities`
`valid_from`	string	Start of validity
`valid_to`	string	End of validity (null = still true)
`tier`	string	Kind of claim
`evidence`	string	What it rests on
`confidence`	string	high \| medium \| low
`source_id`	string	Source → `sources`
`work_id`	string	Cited work → `entities`
`locator`	string	Issue, page, timestamp

Predicates 80 rows

The controlled vocabulary: every relation and attribute the Codex can state.

predicates.csv · About this table

Columns

Column	Type	Meaning
`name`	string	Predicate
`datatype`	string	text \| number \| entity \| color \| date \| bool
`layer`	string	fact \| relation \| spec \| constraint \| community \| meta
`description`	string	Meaning

Sources 5 rows

Where rows come from, under what licence.

sources.csv · About this table

Columns

Column	Type	Meaning
`id`	string	Source id
`name`	string	Source
`url`	string	Source URL
`licence`	string	Licence
`checked_at`	string	When the licence was verified

What is not in the tables yet — and why

The full shape of the database, including the parts we have not filled. A gap listed here is a gap; nothing below is implied to exist.

Table	Holds	Blocked on
`comic_issues`	Every issue: series, number, cover date, on-sale date, price, pages	Grand Comics Database dump (CC BY-SA 4.0). Needs the account owner to log in and accept the GCD data licence.
`issue_credits`	Writer, penciller, inker, colourist, letterer, editor per story	Grand Comics Database dump — same blocker.
`issue_appearances`	Which characters appear in which issue, with first-appearance and death flags	Metron API (CC BY-SA 4.0) for modern issues — needs an account; GCD character fields for the archive.
`story_arcs`	Arcs with ordered issue membership	Metron (arcs are first-class there) — needs an account.
`episodes_seasons`	Full season and episode lists with air dates	Wikidata is partial; TMDB completes it — needs an API key (codex/sources/tmdb.md).
`full_cast_crew`	Complete cast and crew for every production	TMDB — needs an API key. Wikidata cast coverage is real but incomplete.
`box_office`	Budget and gross per production	Wikidata P2130/P2142 — ready to extract, queued.
`creators_bio`	Creator birth/death, nationality, employers	Wikidata person claims — ready to extract, queued.
`anime_manga_detail`	Studios, episode counts, chapter counts, manga-to-anime mapping	AniList and MyAnimeList terms forbid bulk storage — blocked. Wikidata supplies the spine.
`games`	Games, platforms, developers	IGDB terms forbid redistribution without partnership — blocked. Wikidata supplies the spine.
`fan_wiki_infoboxes`	Height, weight, eyes, hair, base of operations, full team rosters	Fandom (Marvel/DC Database) — blocked: its Terms of Use forbid automated access without written permission. Parser is built and tested offline.

Last updated 2026-09-06