Redis Guardrails and Playbooks

Purpose

Help an assistant propose safe, production-ready Redis patterns and operational guidance while avoiding high-risk mistakes.

When to use

Use this skill when the user asks about:

• Using Redis as a cache, session store, rate limiter, lock provider, queue/stream, or pub/sub
• Choosing TTL, eviction, key naming, or data modeling in Redis
• Operating Redis safely (scanning keys, deleting keys, mitigating hot keys, memory issues)
• Migrating or scaling Redis (single node → replicas → cluster)

Avoid this skill when:

• The user needs vendor-specific operational runbooks (managed service dashboards, exact console clicks)
• The request is unrelated to Redis

Inputs

• Scenario: what Redis is being used for (cache, sessions, rate limiting, locks, queues, pub/sub, etc.)
• Constraints: latency requirements, consistency needs, scale/QPS, key cardinality, data size, and deployment mode (standalone/replica/cluster)

If any of the above are missing or ambiguous, ask targeted questions before recommending a design.

Outputs

• A structured playbook (Markdown) that includes:
  • recommended approach and data layout
  • key naming scheme and TTL strategy
  • operational guardrails (safe commands and anti-patterns)
  • failure modes and mitigation
  • verification and rollback guidance

Steps

1. Clarify the scenario and constraints.
2. Identify the Redis deployment mode (standalone/replica/cluster) and persistence requirements.
3. Choose the closest playbook template and tailor it:
   • cache
   • sessions
   • rate limiting
   • distributed locks
   • queues/streams
4. Produce a safe design:
   • key naming and namespacing
   • TTL strategy (what should expire, when, and why)
   • read/write patterns and concurrency controls
5. Add guardrails:
   • prefer SCAN over KEYS for key discovery
   • avoid mass deletes without a plan (pipeline/batches)
   • call out cluster pitfalls (multi-key ops, hash tags)
6. Provide verification steps:
   • functional tests
   • load/latency checks
   • operational signals (memory, eviction, slowlog)
7. Provide a rollback or mitigation plan.

Verification

• The recommendation matches the stated scenario and constraints
• Key naming and TTL strategy are explicit
• High-risk commands are either avoided or gated behind explicit user approval
• Cluster considerations are addressed when relevant (hash tags, multi-key ops)
• Failure modes and mitigation/rollback steps are documented

Boundaries

• MUST NOT recommend KEYS * on production datasets; use SCAN and batch operations
• MUST NOT recommend FLUSHALL/FLUSHDB or mass deletion without explicit approval and an environment check
• MUST NOT suggest disabling persistence, changing eviction policy, or resizing memory without confirming impact
• MUST NOT claim distributed locking is "solved"; clearly state trade-offs and provide safe primitives
• SHOULD NOT propose multi-key operations that break under Redis Cluster without using hash tags or an alternative design

Included assets

• Template: ./templates/redis-playbook.md
• Reference: ./reference/common-pitfalls.md, ./reference/playbooks.md
• Examples: ./examples/