Redis Operations Skills

Redis Operations Skills is a collection of Agent Skills for Redis platform operations, support triage, and production troubleshooting. The repository uses the open Agent Skills layout: each skill is a self-contained directory under skills/ with a required SKILL.md file and optional bundled resources such as scripts/, references/, and assets/.

These skills are distilled operational workflows. They intentionally do not include source knowledge-base articles or full article copies.

Experimental Alpha Status

This repository is experimental alpha-quality material. It is not production-ready operational guidance, and the skills should not be used as the sole authority for Redis Cloud, Redis Software, Kubernetes, billing, security, data deletion, or customer-impacting operations.

Before using any skill output in a real environment, verify the procedure against current Redis documentation, product behavior, support policy, and the live system state. Treat destructive Redis, cloud, billing, account, or data operations as confirmation-sensitive and require a qualified human operator to approve them.

Repository Layout

.
├── skills/
│   └── <skill-name>/
│       ├── SKILL.md
│       ├── scripts/      # optional executable helpers
│       ├── references/   # optional detailed reference material
│       └── assets/       # optional templates or reusable assets
└── scripts/
    └── validate-skills.sh

At setup time, this repository contains 246 skill bundles.

Topic Areas

The skills cover Redis operations across these areas:

• Redis Cloud operations: account setup, database creation, connection workflows, backups, alerts, monitoring, failover tests, maintenance notifications, API usage, Terraform behavior, version policy, Flex, Essentials, Pro, and platform orientation.
• Redis Cloud billing and marketplace workflows: invoices, payment failures, credit-card retries, cost reports, network transfer charges, post-cancellation billing, plan-change billing, AWS Marketplace, GCP Marketplace, private offers, ownership, permissions, and account deletion.
• Redis Cloud security and access: RBAC, Data ACLs, credential rotation, TLS, IP allow lists, VPC peering, private networking, authentication errors, account recovery, team access, SAML/SSO, and Okta troubleshooting.
• Redis Software operations: installation, cluster deployment, licensing, DNS, firewalls, storage, persistence, logging, auditing, monitoring, support packages, node maintenance, upgrades, rollback, certificate management, and cluster recovery.
• Redis Enterprise for Kubernetes and OpenShift: REC/REDB licensing, FQDN discovery, rack awareness, reconciliation behavior, degraded recovery, pod restart triage, secret/auth changes, TLS failures, Istio ingress, VKS support, and OpenShift-specific troubleshooting.
• Active-Active and CRDB: CRDB creation, resize, shard scaling, replication link failures, sync stalls, tombstone memory, credential sync, multi-key command guardrails, migration, flush safety, and participant reachability.
• Search, Query Engine, Redis 8, and vector workloads: index creation, document ingestion, FT.EXPLAIN, FT.PROFILE, FT.DROPINDEX DD, Redis 8 command deprecations, ACL category migration, RESP compatibility, JSON modeling, numeric/date precision, vector search architecture, RedisVL-oriented design, and faceted search.
• Performance, capacity, and reliability: high memory and high network usage, throughput sizing, right-sizing, hot keys, big keys, large keys, bulk deletion, slow log and SCAN diagnostics, connection surges, reconnect storms, Pub/Sub load, endpoint rebinding, failover validation, and latency reduction.
• Data movement and recovery: imports, large dataset validation, migration options, OSS-to-Redis Software migration, Sidekiq and CRDB migrations, shard-count reduction, AOF recovery after FLUSHDB, backup/restore workflows, and safe deletion.
• Client and tool workflows: Redis Insight onboarding, connection, DNS, data exploration, and performance monitoring; redis-cli on Windows through Docker or WSL; support-ticket lookup; and product support routing for Redis Cloud, Redis Software, Redis Open Source, Query Engine, Kubernetes, and Redis Insight.

Using These Skills

Use the skills/ directory as the canonical source of skill bundles.

For agents that can load an arbitrary skills directory, point the agent at skills/.

For Codex user-scope installation, copy or symlink selected skill directories into $CODEX_HOME/skills or ~/.codex/skills when CODEX_HOME is unset.

For Codex repository-scope installation, copy or symlink selected skill directories into the target repository's .agents/skills/ directory.

For Claude or other Agent Skills-compatible clients, install, upload, or package the desired skill directories according to that client's current skill-loading mechanism.

Skill Authoring Conventions

• Keep one skill per focused operational workflow.
• Keep SKILL.md concise and action-oriented.
• Put required trigger coverage in the frontmatter description; agents use that field for discovery.
• Use references/ only for detailed material that should be loaded conditionally.
• Use scripts/ for repeatable checks or fragile procedures that benefit from deterministic execution.
• Do not copy full source articles into skills.
• Do not hard-code current pricing, cloud-provider support windows, marketplace behavior, or version support without live verification guidance.
• Treat destructive Redis, cloud, billing, and account operations as confirmation-sensitive.

Skill Evals

The evals/ directory contains the current harness for comparing selected Redis operations skills against no-skill baselines. It currently covers the first four evaled skills: Cloud database deletion, Cloud subscription cancellation, Kubernetes secret/auth connection failures, and Redis Software node maintenance patching.

See evals/README.md for the current progress snapshot and the skill-structure lessons learned from improving those skills.

Validation

Validate all skill bundles after editing:

./scripts/validate-skills.sh

The script uses SKILL_VALIDATOR when set. Otherwise, it looks for Codex's local quick_validate.py from the skill-creator skill:

SKILL_VALIDATOR=/path/to/quick_validate.py ./scripts/validate-skills.sh

Validation should pass before publishing or installing updated skills.