Skip to main content
By the end of this guide you will have Cache-Pot running locally, a Redis client talking to it, a working vector search, and a semantic cache query — plus a live look at the built-in web dashboard. The whole walkthrough takes less than five minutes from a fresh machine.
1

Install Cache-Pot

Pick whichever method suits your environment. Both options produce the same running server.
go install github.com/subh05sus/cache-pot/cmd/cache-pot@latest
The go install path requires Go 1.25 or newer and places the binary in $GOPATH/bin. The Docker image requires no local Go toolchain.
2

Start the server

If you installed via Go, launch the binary directly. You should see two startup lines confirming the RESP2 listener and the dashboard are both up.
cache-pot

cache-pot: listening on [::]:6379
cache-pot: dashboard on http://localhost:8080
Cache-Pot listens for Redis clients on :6379 and serves the management dashboard on :8080. Both ports are configurable — see the configuration reference for details.
3

Connect with redis-cli

Open a second terminal and connect with redis-cli. Any Redis client or library works in exactly the same way — this step just confirms the server is healthy.
redis-cli ping

PONG
Now set and retrieve a key:
redis-cli set hello world
# OK

redis-cli get hello
# "world"
Any Redis client library works with Cache-Pot out of the box. Here are quick examples for two popular ones.
const Redis = require('ioredis');
const client = new Redis({ host: 'localhost', port: 6379 });
await client.set('hello', 'world');
const val = await client.get('hello');
console.log(val); // world
4

Try the AI features

Vector search stores your embeddings directly — pass the float values and Cache-Pot handles the index. No API key is needed for vector search itself.
# Insert two vectors into the "embeddings" collection
redis-cli VSET embeddings doc1 0.1 0.2 0.3 META "hello world"
redis-cli VSET embeddings doc2 0.9 0.8 0.7 META "goodbye world"

# Find the nearest neighbour to a query vector
redis-cli VSEARCH embeddings 0.1 0.2 0.3
# 1) "doc1"
# 2) "hello world"
Semantic caching stores model responses and returns them when a close-enough prompt arrives later, saving you a model call.
redis-cli SCACHE.SET "What is the capital of France?" "Paris"
redis-cli SCACHE.GET "whats the capital of france" THRESHOLD 0.9
# "Paris"
Semantic caching requires an embeddings provider. Set CACHEPOT_EMBED_URL, CACHEPOT_EMBED_MODEL, and (if needed) CACHEPOT_EMBED_KEY before starting the server. A free local Ollama instance works, and so does the OpenAI embeddings API. See Semantic Cache configuration for setup instructions.
5

Open the dashboard

Navigate to http://localhost:8080 in your browser. The dashboard is baked into the binary — no build step, no external assets.From the dashboard you can:
  • Watch live command throughput, memory usage, and key counts on the Overview panel
  • Browse, inspect, and edit every key in the Browser
  • Run commands interactively in the Workbench
  • Stream a live command log with the Profiler
  • Review slow commands in the SlowLog
  • Publish and subscribe to channels in the Pub/Sub panel
  • Analyse memory distribution by type and namespace in Analysis
  • List and kill active connections from Clients

​
What’s next

You now have a fully working Cache-Pot instance. Explore the areas below to go deeper.
  • Introduction — understand the full feature set and how Cache-Pot compares to Redis
  • Installation — persistent volumes, build-from-source, and production flags
  • Configuration — authentication, persistence (snapshots and AOF), and embedding providers
  • AI Agents & MCP — connect Claude or another MCP agent to Cache-Pot