redis agent-skills
OFFICIALLABSCO SUMMARY
Six cover the parts of Redis that apply whether it is self-hosted or a managed service: redis-core (choosing between String, Hash, List, Set, Sorted Set, JSON, Stream, or Vector Set, and consistent key naming), redis-connections (pooling, pipelining, client-side caching, avoiding slow commands like KEYS and HGETALL), redis-search (FT.CREATE schema design, vector similarity with HNSW, hybrid lexical-plus-vector retrieval for RAG), redis-clustering (hash tags, avoiding CROSSSLOT errors, reading from replicas), redis-security (ACLs, TLS, network exposure), and redis-observability (SLOWLOG, MEMORY DOCTOR, FT.PROFILE, wiring metrics into Prometheus or Datadog). The other two are specific to Redis's own hosted products rather than Redis the database: redis-semantic-cache covers LangCache, a semantic cache for LLM responses that runs on Redis Cloud, and iris-development covers Iris, Redis's agent-memory data plane, also on Redis Cloud, using its own redis-agent-memory and @redis-iris/agent-memory SDKs.
This is for engineers already running Redis who want an agent that reaches for the right data structure and command instead of a generic key-value pattern; the six general skills work against any Redis deployment, but the LLM-caching and agent-memory pair only apply if you are also on Redis Cloud specifically. The repository installs three ways -- the Agent Skills CLI, a Claude Code plugin, or a Cursor plugin -- and says the plugin folders symlink into the same top-level skills/ directory rather than forking it, so there is one source of truth regardless of which install path you take.
READ THE FULL ANALYSIS
What it costs to start. Six of the eight skills need nothing but a Redis connection you already have. The remaining two are labeled account_key: LangCache and Iris are Redis Cloud features, so redis-semantic-cache and iris-development do nothing against a plain open-source Redis instance.
Checked from the package's own README and its eight skill descriptions.
ALSO IN THIS PACKAGE
redis-development
Redis development best practices — data structures, query engine, vector search, caching, and performance optimization
WHAT'S INSIDE
8 showing · 8 totalNothing else to set up — install it and go.
iris-development
Gives an AI agent a memory: what was said in the conversation it is having now, plus a searchable store of facts it can pull back weeks later — hosted by Redis rather than run by you.
redis-clustering
Once a Redis database is spread over several machines, a command that touches two pieces of data at once fails unless those two were deliberately put on the same machine. This is how to name things so that stops happening, and how to send read traffic to the spare copies instead of the busy one.
redis-connections
Most Redis slowness turns out not to be Redis — it is an application opening a fresh connection for every request, asking for everything at once, and then waiting forever when something breaks. These are the habits that avoid all three.
redis-core
Redis offers eight different ways to hold the same information, and the right one depends entirely on how it will be read back later. This is how to choose between them, and how to name what you store so it stays findable.
redis-observability
What to keep an eye on while a Redis server is healthy, and what to reach for the moment it stops being healthy — which numbers actually matter, which built-in command answers which question, and when a screen you can click around in beats a graph.
redis-search
Redis can act as the search engine over the data already sitting in it. This covers setting that up — searching by words, by numbers, by location, and by meaning — and writing queries, including ones that weigh wording and meaning at the same time.
redis-security
Locking a Redis server down before it goes live is three separate jobs — proving who is connecting, limiting what each one is allowed to do, and keeping the server off the open internet. This treats them as one piece of work rather than three.
redis-semantic-cache
When somebody asks a question close enough to one already answered, hand back the stored answer instead of paying an AI model to write it again. Setting that up on Redis's hosted cache is what this covers.
HOW TO GET IT
npx skills add redis/agent-skillsnpx skills add redis/agent-skills --skill <name> --full-depthPick the skill name from the Skills tab — each entry there installs independently.
/plugin marketplace add redis/agent-skills/plugin install redis-development@redisTyped inside the agent's own prompt, not in a terminal. The marketplace is called redis, which is the part after the @.