redis/agent-skills
10 skills3.2k installs700 starsGitHub
Install
npx skills add https://github.com/redis/agent-skillsSkills in this repo
1Redis DevelopmentRedis Development is an MIT-licensed agent skill from Redis that packages a full performance and modeling playbook for Redis workloads. Solo and indie builders shipping APIs, session stores, real-time features, or RAG pipelines install it when agents need consistent guidance on key design, TTLs, connections, JSON documents, Redis Query Engine indexes, RedisVL vector search, and LangCache semantic caching. The skill organizes 29 rules in eleven categories with explicit priority tiers so automated refactors and greenfield code generation favor high-impact fixes first. It is not a Redis installer or hosted service—it is procedural knowledge meant to sit beside your stack docs while you design caches, tune memory, or debug slow FT.SEARCH queries. Use it during backend implementation, before launch perf passes, and when operating production Redis where memory and replication matter.3.2kinstalls2Iris Developmentiris-development teaches agents how to integrate Redis Iris—the Redis Agent Memory service—using the official redis-agent-memory (Python) and @redis-iris/agent-memory (TypeScript) SDKs. Solo builders shipping assistants or multi-step agents use it when memory must survive sessions: append timeline events, promote facts to searchable long-term storage, and query what the user already told the system. The skill emphasizes correct bulk ingestion patterns, capped at one hundred memories per request, with stable IDs you control so retries stay safe. It also covers authentication, service provisioning, and the promotion workflow between short-lived session state and durable memory. This is implementation-focused documentation packaged as a skill, not a hosted MCP server. Expect intermediate familiarity with Redis Cloud or your Iris endpoint and with async agent loops that need deterministic memory upserts.1installs3Redis Clusteringredis-clustering is a Redis-maintained agent skill that teaches how Redis Cluster hashes keys across slots and how to keep related keys together using hash tags. Solo builders shipping APIs or SaaS backends often hit CROSSSLOT failures when they pipeline or run multi-key commands without planning key names. The skill walks through correct patterns in redis-py and Jedis—setting namespaced keys with a shared tag, executing pipelines, and using multi-key commands like LMOVE on co-located keys. It also frames replication and replica reads as part of durability and read-scaling choices. Use it while operating clustered Redis in production or while designing backend data access during build so agents do not suggest illegal cross-slot operations.1installs4Redis Connectionsredis-connections is an official-style Redis agent skill aimed at solo builders who add caching, queues, or session stores without a dedicated platform team. It packages operational wisdom—how to open and reuse connections, when to pipeline, how client-side caching fits in, and which commands block the server at scale—into guidance your agent can apply while writing or reviewing code. The readme’s production table is the anchor: never iterate with KEYS, prefer SCAN with cursors, and paginate large list reads. Language snippets span Python and Java, which helps when your stack is not Node-only. Use during backend implementation and again when production latency spikes or Redis CPU pegs, because the same rules apply at ship and operate time. It is not a full Redis schema or module tutorial; it keeps the agent focused on connection behavior and safe access patterns so small apps do not take down a shared instance with one hot path.1installs5Redis CoreRedis Core is an agent skill published by Redis for solo builders adding caching, session stores, rate limits, or real-time feeds without reinventing data-modeling rules. It compresses foundational guidance—how to name keys, set TTLs, choose Hash versus JSON types, and decide between Streams and Pub/Sub—into procedural knowledge your coding agent can apply while you implement backend features. The skill ships as part of Redis's evaluated agent-skills repository, where benchmark runs compare agent performance with and without the skill on core modeling scenarios. Reach for it during Build when sketching access patterns, and again in Operate when debugging memory pressure or hot keys that trace back to weak naming or missing expiration.1installs6Redis ObservabilityRedis observability teaches you how to monitor and debug Redis instances using built-in commands like INFO, SLOWLOG, MEMORY DOCTOR, and FT.PROFILE. Use this skill when your Redis deployment is running in production and you need to identify performance bottlenecks, memory leaks, or connection issues. It matters because proactive monitoring prevents cache misses, connection rejections, and slow query cascades from impacting your application.1installs7Redis Query Engineredis-query-engine is an agent skill that packages procedural knowledge for Redis Query Engine: building FT.CREATE indexes, querying with FT.SEARCH and FT.AGGREGATE, and standardizing on DIALECT 2. Solo builders shipping APIs, SaaS catalogs, or agent tools backed by Redis use it when full-text or structured search must live beside primary key-value data without bolting on a separate search cluster. The skill emphasizes explicit DIALECT 2 in raw commands or via client defaults so special characters, NULL values, and vector queries parse predictably—dialects 1, 3, and 4 are called out as deprecated in Redis 8. Examples span Python redis-py and raw Redis command form, which helps whether you automate indexing in app code or validate queries in redis-cli. It also surfaces indexing keywords from the upstream package (ft-create, ft-search, query-engine) and zero-downtime update guidance for teams that cannot afford search downtime during schema changes.1installs8Redis Securityredis-security is an agent skill from Redis’s official agent-skills repo that walks solo builders through production Redis hardening: fine-grained ACL users with least privilege, mandatory authentication, TLS, and safer defaults than one shared password on the default user. It is aimed at indie developers shipping APIs or SaaS that depend on Redis for cache, queues, or session storage and need a copy-paste security baseline before go-live. Use it when you are configuring a new production instance, auditing an existing deployment, or fixing overly broad requirepass access. The skill emphasizes concrete ACL recipes (read-only on key patterns, writers without dangerous commands) and links to Redis ACL documentation so your agent can align redis.conf and client code with the same policy.1installs9Redis Semantic Cacheredis-semantic-cache teaches agents how to configure Redis LangCache for semantic caching of LLM responses: initialize the client, set similarity thresholds to balance false positives versus hit rate, and split workloads across dedicated cache IDs so support chat does not collide with code-completion traffic. Solo builders shipping agent features or API wrappers can cut latency and token spend when prompts are near-duplicates, as long as they accept preview-service constraints on Redis Cloud. Use this skill when you already have Redis Cloud LangCache provisioned and need procedural wiring—not when you only need in-memory dict caching. Pair it with your existing LLM route so search-then-generate follows a proper cache-aside pattern.1installs10Redis Vector SearchRedis Vector Search is an official Redis agent skill that teaches coding agents how to stand up embedding indexes in Redis and query them with RedisVL for RAG and semantic search. Solo builders shipping chatbots, copilots, or search-over-docs features use it when they need a concrete answer on HNSW versus FLAT, distance metrics, dimensionality, and index tuning instead of copying fragmented Stack Overflow snippets. The material walks through schema definition, algorithm choice, and hybrid retrieval patterns so retrieval latency and recall targets are explicit before production traffic hits. It fits indie teams that already chose Redis for caching or queues and want one datastore for vectors without operating a separate vector database on day one. Complexity is intermediate because you must understand embeddings shape, recall tuning, and operational memory costs. Pair it with your embedding provider skill and application backend work; it does not replace observability or security review for exposed Redis endpoints.1installs