Developer builds Kin middleware to centralize AI app account access

On November 3, 2026, a developer published a post in the AI section of DEV Community describing a middle layer they built themselves, called Kin. The author explains that they use models from four labs every week, and every one of them wanted accounts connected inside its own app, each with its own idea of what was allowed. They could not see what any of them did with that access, and there was no single switch to stop them, so they put a layer in the middle that they control. The author also notes that they built one others can use, called Kin. According to the post, the design is straightforward: connect everything to the layer once. The layer holds every connection, whether MCP servers, OpenAPI endpoints, or local tools on the machine. Users sign in once, and AI applications never hold tokens directly; they go through the layer to reach whatever account or tool they need. The author adds that Kin also offers a small fixed set of tools, which reduces the burden on the model's context window compared with exposing many raw capabilities at once. At this stage Kin is known mainly through the author's own write-up, with no other sources reporting on it yet. What the post presents is a personal account of a self-built solution rather than an announced product release, and the described motivation is lack of visibility and control over how AI applications use account connections.