Skip to main content
Cache-Pot is an in-memory data store built in the Redis mould and reworked for the way AI applications actually use a cache. It speaks the Redis wire protocol out of the box, so every client and library you already rely on keeps working without modification — while also giving you vector search, semantic response caching, and a native MCP endpoint for AI agents, all packed into a single self-contained binary.

​
Where to begin

Introduction

Understand what Cache-Pot is, how it differs from Redis, and when to reach for it.

Quickstart

Install Cache-Pot, connect a Redis client, and run your first vector search in under 5 minutes.

AI Agents & MCP

Wire Cache-Pot into Claude or any MCP-compatible agent as a first-class tool.

Vector Commands

Full reference for VSET, VSEARCH, VDEL, VCARD, and VDIM.

​
Get up and running in three steps

1

Install Cache-Pot

Grab the binary with a single go install command or pull the Docker image — no external dependencies required.
go install github.com/subh05sus/cache-pot/cmd/cache-pot@latest
2

Start the server

Run the binary. Cache-Pot listens for Redis clients on :6379 and opens a web dashboard on :8080.
cache-pot
# cache-pot: listening on [::]:6379
# cache-pot: dashboard on http://localhost:8080
3

Connect and use it

Point any Redis client or redis-cli at localhost:6379 and start issuing commands. Switch to AI features — semantic caching, vector search, agent memory — whenever you need them.
redis-cli ping
# PONG
redis-cli set hello world
# OK
redis-cli get hello
# "world"

​
Three capabilities in one binary

Redis-compatible

Cache-Pot speaks RESP2, the same wire protocol your Redis clients already use. Strings, hashes, lists, sets, sorted sets, pub/sub, transactions, TTLs — all the everyday commands work without changing a line of application code.

AI-native

Vector search and semantic caching ship built-in. Store and query high-dimensional embeddings with VSET/VSEARCH, or cache model responses by meaning with SCACHE.SET/SCACHE.GET — no extra modules or glue code needed.

MCP endpoint

Cache-Pot includes a native Model Context Protocol server. Connect Claude or any MCP-compatible agent directly, giving it read, write, search, and memory tools with zero adapter setup.

Single binary

The entire stack — cache, vector store, semantic cache, MCP bridge, and web dashboard — lives in one self-contained binary. No external services, no database, no configuration file required to get started.