How to use the Redis Expert agent skill

What is an agent skill and what does redis-expert make the agent do?

An agent skill is a folder containing a SKILL.md file of instructions that an AI agent loads when a task matches it. The redis-expert skill is a Redis guidance document. When it is active, the agent approaches Redis questions as a senior backend engineer specializing in Redis as a data structure server, cache, message broker, and real-time data platform.

In practice the skill shapes the agent's advice. It points toward suitable data types such as sorted sets, hashes, streams and HyperLogLog, and it recommends patterns including cache-aside, distributed locks and sliding-window rate limiters. It also flags pitfalls such as using KEYS in production, storing large blobs in Redis, and relying only on RDB snapshots for durability.

• Picks data types to match an access pattern (for example sorted sets for leaderboards, streams for event logs)
• Recommends patterns like cache-aside, distributed locks and rate limiters
• Warns about common mistakes like unbounded keys, large blobs and snapshot-only persistence

Install the redis-expert skill from the ZIP

Download the redis-expert ZIP from this page. The skill ships as a single file, SKILL.md. There are two ways to install it.

For Claude Code, unzip the download into your user skills folder ~/.claude/skills/ to make it available in all your projects, or into a project's .claude/skills/ folder to limit it to that project. For the Claude apps, upload the ZIP under Settings → Capabilities → Skills.

• Personal install: unzip into ~/.claude/skills/
• Project install: unzip into the project's .claude/skills/ folder
• Claude apps: upload the ZIP under Settings → Capabilities → Skills

How do you trigger the skill with natural requests?

You do not need a special command. The skill is meant to be picked up when your request is about Redis. Describe your access pattern, data or problem in plain language and say Redis is involved, for example "Redis" or the specific command or structure you have in mind.

Good requests include the goal, the data size and the failure you care about. The more concrete you are, the more useful the advice will be.

• "We cache user profiles in Redis. What TTL and invalidation approach should we use?"
• "Design a rate limiter in Redis that allows 100 requests per minute per API key."
• "Review this Lua script for a distributed lock. Is releasing it safe?"
• "Should we store our leaderboard in a sorted set or a hash?"

Examples of what the skill helps you decide

The skill gives reasoned suggestions for specific designs. For a distributed lock, it recommends acquiring with SET lock_key unique_value NX PX 30000 and releasing through a Lua script that checks the value first, so you do not delete another client's lock. For a sliding-window rate limiter, it suggests a sorted set with timestamp scores, ZREMRANGEBYSCORE to prune old entries and ZRANGEBYSCORE to count requests in the window.

For event processing, it points to Streams with XADD, XREADGROUP and consumer groups when messages need acknowledgment, and to Pub/Sub only when losing messages is acceptable. For objects that get partial updates, it suggests hashes with HSET and HGETALL rather than serialized JSON strings.

What does the skill need, and what does it run?

The skill lists no requirements. It does not run scripts and does not use the network. It is written guidance only: it does not connect to any Redis server, so it cannot read your configuration, check your Redis version or inspect your keys.

Because of this, treat its output as a design review, not a verified fact about your system. Confirm commands and settings against the documentation for your own Redis deployment before you apply them, especially anything that touches production data.

• Requirements: none listed
• Runs scripts: no
• Uses network: no
• Included file: SKILL.md

Safety notes and limits

Because the skill only shapes advice, the risk comes from acting on that advice without checking it. Wrong guidance about a live system could affect real data. The skill is not suited to operating or administering a production Redis cluster, where you need to verify your actual configuration and version.

Some of the guidance is about settings you should change with care. For example, it recommends configuring AOF persistence with appendfsync everysec for data that must survive restarts, and setting lua-time-limit so long Lua scripts do not block other clients. Test such changes in a non-production environment first.

• Do not apply configuration changes to production based only on the agent's answer
• Verify commands and settings against your Redis version's documentation
• The skill cannot see your server, so it cannot confirm your current state

Tips for getting better answers

Give the agent the access pattern and the scale: how many keys, how often they are read and written, and what happens when Redis is down. The skill is built to design for graceful degradation, so telling it what your application should do on a Redis outage helps.

Ask it to review code or a design you already have. It is well suited to finding pitfalls such as missing TTLs, KEYS usage and large values in Redis.

• Always include the expected key count and TTL needs
• Ask what should happen when Redis is unavailable
• Paste your Lua script or design and ask for a review