It would be awesome if Readlang could provide an MCP server to let Claude (or whatever) query a user’s vocab and texts of their read articles. For example, after I read an article, Claude could follow up with words I clicked on, coming up with various questions that make me use them in ways other than the flashcards (e.g., new contexts, re-phrasings, more general complete-the-phrase, compare & contrast, …). And in the other direction, it could create new stories or conduct dialogs focusing on words that I’m weak on in general, based on my vocab stats. In addition, it could feed back into Readlang additional words that we discover in these interactions. The list goes on. This would offload the AI expense from Readlang onto the user, while still maintaining Readlang as the central source of all the texts and data, as it is now. I suppose I could hack such a thing by exporting texts and the vocab dumps, but that’s more painful.
Doesn’t it conflict with readlang paid features?
I’m not aware of any Readlang features that do the above? And in general, giving Claude an overview of my vocabulary and texts would open up a whole world of learning beyond what I imagine falls in Readlang’s wheelhouse. It’s again something that could be crudely hacked together with current export and import capabilities, but I think would be a serious value-added if made seamless through mcp.