wtf is cache-pot?

Cache-Pot is an in-memory data store built to work like Redis, but designed around how AI applications and agents actually use a cache. It speaks the same wire protocol as Redis, so an existing Redis client and the code around it keep working untouched once you point it at Cache-Pot instead. What sets it apart from plain Redis is that vector search and a semantic cache are built in rather than added on. You can store vectors and search for the nearest matches directly, which is useful for things like similarity search over documents. The semantic cache lets you save a model's answer to a question once, and later, when a similar but not identical question comes in, Cache-Pot can return the saved answer instead of paying for another call to the model. There is also a native connection for AI agent tools called MCP, so agents such as Claude can read, write, search, and remember through Cache-Pot directly as a tool, without extra glue code. The whole project ships as one self contained binary with no other services to install, and it includes a small web dashboard for watching what is happening inside the store. You can install it with Go, run it through Docker, or build it from source, and once it starts it listens on the usual Redis port as well as a separate port for the dashboard. A built in command line shell lets you talk to it even if you do not have the standard Redis client tools installed. The project is honest about its limits: there is no clustering, replication, or failover, and it is not tuned to beat Redis on raw speed. It is meant as a single machine cache and AI data layer, not a replacement for a large production Redis cluster.