| name | cache-strategy |
| description | Design and implement caching strategies for performance. Invoke for: "add caching", "cache this", "Redis", "cache invalidation", "TTL", "cache miss", "expensive operation", "repeated computation", "memoize", "this is too slow because of repeated calls".
|
| argument-hint | what to cache or cache system to design |
| allowed-tools | Read, Edit, Write, Grep, Glob |
Skill: Cache Strategy — Design & Implement Caching
Category: Development
Role
Design cache layers that dramatically reduce latency and compute cost for repeated operations.
When to invoke
- Repeated expensive operations (DB queries, API calls, computation)
- "cache this response"
- Redis or in-memory caching needed
- Cache invalidation design
Instructions
- Identify: what data? How often changes? How often read? Acceptable staleness?
- Choose cache level: in-process (dict), Redis, CDN, DB query cache
- Design cache key: deterministic, includes all relevant parameters
- Set TTL: based on data freshness requirements
- Design invalidation: time-based? Event-based? Manual?
- Handle: cache miss, stampede (lock), thundering herd
- Implement in src/utils/cache.py
Output format
## Cache Design — <what> — <date>
### Data Profile: reads/writes per minute, staleness tolerance
### Cache Level: in-process / Redis / CDN
### Key Pattern: f"{resource}:{id}:{version}"
### TTL: Xh
### Invalidation: time-based / on-write
### Implementation
Example
/cache-strategy LLM completion responses — cache by (model, prompt_hash, temperature)
