Developers

Endpoints

Lists return every match in one JSON array. There is no pagination.

Filters

• q: Words to match in the title, description, topic or word list (also `query` or `search`).
• level: easy, challenge or expert. Each list's OpenAPI entry names the levels it has.
• category: A topic slug such as animals. Word searches, criss-cross, word scrambles, spot the difference, crosswords, nonograms, and bingo only.

Examples

curl https://professorotter.com/api/v1/catalog
curl "https://professorotter.com/api/v1/catalog/word-searches?q=farm&level=easy"
curl https://professorotter.com/api/v1/catalog/products
curl -H "Accept: text/markdown" https://professorotter.com/api/v1/catalog/mazes
curl https://professorotter.com/api/v1/catalog/products.md
curl -i https://professorotter.com/api/v1/catalog/mazes/not-a-maze

JSON and Markdown

Responses are JSON unless you ask for Markdown with Accept: text/markdown or by adding .md to the path. Every public page on the site works the same way: /shop.md is the shop as Markdown.

Errors

Errors are RFC 9457 problem details with application/problem+json. code is stable, and resolution says what to do next.

{
  "type": "about:blank",
  "title": "Not Found",
  "status": 404,
  "detail": "No catalog collection matches /api/v1/catalog/mazes/not-a-maze.",
  "code": "not_found",
  "resolution": "Check the path against /llms.txt or GET /api/v1/catalog."
}

Versioning and deprecation

Canonical paths start with /api/v1/catalog. The unversioned /api/catalog paths are aliases of v1 and return the same responses.

Breaking changes ship as a new version path (/api/v2/catalog); adding fields or collections is not breaking and ships in v1. Nothing is deprecated today. When a version or the unversioned /api/catalog alias is deprecated, its responses carry a Deprecation header (RFC 9745), a Sunset header (RFC 8594) with the removal date, and a Link header with rel="deprecation" pointing at /developers, at least 90 days before it stops answering. The notice is also posted on /developers and in this document.

Discovery files

• OpenAPI 3.1
• llms.txt
• llms-full.txt
• RFC 9727 API catalog
• ARD catalog
• Agent Skills index
• Markdown sitemap

MCP server

• URL: https://professorotter.com/mcp
• Server card: https://professorotter.com/.well-known/mcp/server-card.json

Read-only Streamable HTTP tools, no sign-in: find activities, get one with its PDF link, list the shop, and play a word search right in the chat. make_word_search turns a child's spelling list into a word search to play and print. Hosts that support MCP Apps show the grid as a game. In Claude: Settings > Connectors > Add custom connector, then paste https://professorotter.com/mcp.

WebMCP

In a browser with WebMCP (`document.modelContext`, a W3C Community Group draft), these pages offer tools to the browser's agent. Other browsers load nothing extra and see no change.

• make_word_search on https://professorotter.com/word-search-maker: Make a word search on this page from a list of 4 to 20 words, like a child's spelling list. It fills in the maker form, shows the puzzle on the page, and returns the grid, the words, a link to share the puzzle (playUrl) and a link to print it as a PDF with an answer key (pdfUrl). If the list breaks a rule, the result's error says what to fix.
• print_this_puzzle on https://professorotter.com/word-searches/{slug}: Get the printable PDF of the word search on this page, in the layout shown now, with an answer key on the second page. Returns pdfUrl for the person to open and print; it doesn't open anything itself.

Buying

Checkout is not an API. A person buys in the shop through Stripe Checkout. Agents must never invent Stripe keys or charge a card. Prices are on the pricing page.

Contact

Questions about the API go to hello@professorotter.com.