Shelf map · pass №1 · 2026-09-04

The Claude connectors directory, by who each connector is for

The directory's counter says how many connectors there are. It does not say who any of them is for. This page does: every card in the categories my agents walked carries one label, customers, staff, developers, platforms or unclear, and the table is rendered from the dataset, not typed. On 2026-09-04 the whole directory held 2,383 connectors. Of the 170 cards labelled so far, 18 face a customer at all.

CategoryWalkedCustomersStaffDevelopersPlatformsUnclear
Health27 of 27 · whole1220103
Life sciences44 of 44 · whole1195181
Legal99 of 99 · whole5621301
17018836585
Pass №1 · 2026-09-04. «Customers» means a customer of that business can do something with that business. «Platforms» are marketplaces, aggregators and public reference sources, which is to say not one business. «Unclear» is reported as its own number and never folded into another bucket.

What the Claude connectors directory does not show

The counter reads as a crowded market. Stand on the customer's side of the counter and it thins out fast: of 18 customer-facing cards, 6 belong to a business doing a service for named clients rather than to a software product whose users are its customers. The rest of the directory serves people inside companies, serves engineers, or serves the market at large, and 5 cards do not say enough to be labelled at all.

  • Health: 27 cards, walked whole. 12 face a customer; 3 are a business answering its own clients.
  • Life sciences: 44 cards, walked whole. 1 face a customer; none are a business answering its own clients.
  • Legal: 99 cards, walked whole. 5 face a customer; 3 are a business answering its own clients.

What each shelf holds

Each label describes the person on the other end of the connector, read from nothing but the card's own name and description.

  • Customers: serves the business's own customers. A studio answering about its own classes; an app letting its users read their own data. Within this shelf the dataset flags the second reading separately: a business serving named clients, as opposed to a product serving its users.
  • Staff: serves people working inside a company. A contract tool for the legal team, a lab notebook for the researchers, a data room for the deal team.
  • Developers: serves engineers.
  • Platforms: serves many businesses at once, or is a public reference source. A court-records search, a literature database.
  • Unclear: the card does not say. Reported as its own number and never folded into another bucket.

How the count is made

The directory shows its cards only to a logged-in profile, and the platform's terms rule out automating that page, so the one step a person does is open each category in a browser and scroll it to the end. A snippet my agents wrote reads every card as the directory rendered it: name, description, and whether it sits in the community section. Everything after that, the dataset, the labels, the table, is their work. A label reads the card, not the product behind it; where the card does not say who it serves, the label is unclear and stays unclear. A whole walk is a whole walk on the day, which is why every pass carries a date and an ordinal.

The dataset

Everything on this page renders from one file: /data/shelf-map.json, CC BY 4.0. It holds the labelling rule, every card and every label. A reader who thinks a label is wrong can name the card and say why, which is the point of publishing the labels rather than the totals.

Read next