3 connectors, live from D1
Connectors
MCP concepts, known pre-built connectors, and this repo's own dogfooded naming pattern, grouped by category unless filtered.
| connector | real/pattern | description |
|---|---|---|
| Connectors vs. custom tools | real | A "connector" is typically a pre-built, ready-to-use integration (often maintained by Anthropic or a third party) that wraps an MCP server for a specific external product, so a user can enable it without writing any code. A "custom tool" is a tool an application or developer defines and registers directly (for example, in the Managed Agents API or via a hand-written MCP server) — it is not necessarily wrapped in a pre-built connector UI, and it can be scoped to exactly the actions an application wants to allow. |
| Connectors are one instance of a broader MCP-server ecosystem | real | Beyond the handful of well-known pre-built connectors, Claude can connect to any MCP server a user or organization stands up themselves — including this repo's own filesystem, Cloudflare, Postgres, and engineering-coworker MCP servers, none of which are "pre-built connectors" in the marketed sense but which use the identical MCP transport and tool-call mechanism. |
| Model Context Protocol (MCP) | real | MCP is an open protocol that lets an AI assistant like Claude connect to external data sources and tools through a standardized server interface, rather than each integration being bespoke. An MCP server exposes a set of tools (and sometimes resources/prompts) over a defined transport (e.g. stdio or HTTP), and any MCP-compatible client — Claude Desktop, Claude Code, Claude cloud products — can connect to it using the same protocol. |
Machine-readable version: GET /api/connectors?category=concept