Return the authenticated World Anvil user (id + username). Use to verify credentials are working.

List all worlds owned by (or shared with) the authenticated user, with their ids. Usually the first call in a session — most other tools need a world_id from here.

Full details of one world: description, settings, counts and links. Use list_worlds first to find the world_id.

List ALL articles in a world (auto-paginates; nothing is hidden). Every World Anvil entity — NPCs (person), locations, organizations, session reports, plots, myths — is an article with a template type, shown as [type] on each line. Optionally filter by category or template type. For keyword lookups prefer search_articles.

Find articles in a world by keyword. Case-insensitive match against title, slug and tags. Multiple words must all match. Fast (metadata only) — to search inside article body text use search_content instead. Use get_article on a result id to read content.

Read one article in full: content (BBCode), sidebar, tags, category, relations, and template-specific fields (a person's dob/ethnicity/organizations, etc.). Content uses World Anvil BBCode; links to other articles appear as @Title mentions — those uuids can be fetched with get_article too, which makes this the way to walk the world's knowledge graph for context.

Create a new article in a world. Content is World Anvil BBCode ([h1]..[/h1], [b], [url], @Title mentions). New articles default to private+draft so you can review before publishing (pass state='public' and is_draft=false to publish immediately).

Update an existing article. Only the fields you provide are changed. NOTE: content is REPLACED wholesale — to edit, get_article first, modify, send back the full body.

PERMANENTLY delete an article. This is irreversible — World Anvil has no trash bin for articles. Confirm with the user before deleting anything they didn't explicitly name.

List all categories (folders) of a world. Articles reference these by id in their category= field; use the ids to filter list_articles or to file new articles.

Create a category (folder) in a world, optionally nested under a parent category.

Rename a category or move it under a different parent. Only provided fields change.

PERMANENTLY delete a category (irreversible). Articles inside are NOT deleted — they become uncategorized. Confirm with the user first.

Full-text search INSIDE the world's text: article bodies (content, sidebars, footnotes …), GM-only secrets, and notebook notes, returning matches with context snippets. All words must appear in the same entity (case-insensitive). Backed by a local cache that self-refreshes on every call, so results are always current; the very first call on a world builds the cache and can take a minute. For title/tag lookups search_articles is faster.

Find every article, secret or note that LINKS TO a given article. World Anvil stores links as @Title mentions, so this searches cached text for the target's uuid — exact backlinks, no false positives from name coincidences. Use it for questions like 'where is this NPC mentioned?' or before deleting an article. Self-refreshing cache; first call on a world builds it (can take a minute).

List a world's non-article resources: secrets (GM-only notes), maps, timelines, histories (timeline events), notebooks, or subscribergroups (player visibility groups). Auto-paginates; returns one line per asset with its id. Use get_asset to read one in full. For articles use list_articles instead.

Read a single asset in full. kind=map returns the ASSEMBLED hierarchy (map + layers + marker groups + all markers with their article links — several API calls, worth it: markers are where geography meets lore). kind=notebook returns notebook + sections + notes. secret/timeline/history/subscribergroup return the entity itself. For articles use get_article.

Clone a COMPLETE world to a local directory as raw full-granularity JSON — articles, categories, secrets, map hierarchies (layers/groups/markers), timelines, histories, notebooks+notes, subscriber groups, world meta — plus downloaded image binaries and an entities.json id→{title,slug,file} index for rewriting @mention links during migration. Makes hundreds of API calls; takes minutes on a large world. Returns a summary; data goes to disk, NOT into context — read specific exported files afterwards if needed. Refuses to write into a non-empty directory that isn't a previous export. Requires an absolute output path.