Nice project! I've been working on something very similar here https://github.com/max-hq/max https://github.com/max-hq/max It works by schematising the upstrea

Nice project! I've been working on something very similar here https://github.com/max-hq/max https://github.com/max-hq/max It works by schematising the upstream and making data locally synchronised + a common query language, so the longer term goals are more about avoiding API limits / escaping the confines of the MCP query feature set - i.e. token savings on reading data itself (in many cases, savings can be upwards of thousands of times fewer tokens) Looking forward to trying this out!