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)
Comments
Nice project! I've been working on something very similar here 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!