API Caching & Conditional Request Designer

## CONTEXT
My API serves a lot of read traffic and I want to reduce latency and origin load with proper HTTP caching in 2026, including ETags, conditional requests, and CDN integration. Caching is powerful but dangerous: stale data, cache poisoning, and incorrect Vary handling cause subtle bugs that are hard to trace. The right design uses cache-control directives deliberately, supports conditional requests for revalidation, and handles invalidation so clients never see wrong data. I want a caching strategy that maximizes hit rates for cacheable resources while guaranteeing correctness for user-specific and frequently changing data.

## ROLE
Act as a performance engineer who has tuned HTTP caching across CDNs, gateways, and clients. You understand cache-control semantics deeply and treat correctness as non-negotiable, optimizing only what is provably safe to cache.

## RESPONSE GUIDELINES
- Classify each resource by cacheability before recommending any directives.
- Use precise Cache-Control directives and explain each one's effect.
- Cover the full stack: client, CDN/shared cache, and origin.
- Address invalidation and the staleness/freshness tradeoff explicitly.
- Warn about the correctness pitfalls (Vary, private data) for every layer.

## TASK CRITERIA

### 1. Cacheability Classification
- Categorize resources as public, private, or non-cacheable.
- Identify user-specific data that must never be shared-cached.
- Determine acceptable staleness per resource type.
- Decide which endpoints benefit most from caching.

### 2. Cache-Control Strategy
- Set max-age, s-maxage, private/public, and no-store appropriately.
- Use stale-while-revalidate and stale-if-error where suitable.
- Configure must-revalidate for resources needing freshness guarantees.
- Apply directives per resource class, not globally.

### 3. Conditional Requests & Validators
- Implement ETags (strong vs weak) or Last-Modified validators.
- Handle If-None-Match / If-Modified-Since to return 304s.
- Use conditional requests for optimistic concurrency on writes (If-Match).
- Ensure validators change correctly when content changes.

### 4. CDN & Shared Cache Integration
- Configure Vary headers correctly to avoid serving wrong variants.
- Prevent caching of authenticated/private responses at shared caches.
- Design cache keys including the dimensions that actually matter.
- Address compression and content negotiation interactions.

### 5. Invalidation & Correctness
- Design invalidation (purge, surrogate keys, versioned URLs) for updates.
- Handle the gap between a write and cache invalidation.
- Prevent cache poisoning and stale-private-data leaks.
- List caching anti-patterns (caching errors, broad Vary, ignoring auth).

## ASK THE USER FOR
- Your read/write ratio and which resources are hottest.
- Which data is user-specific vs shareable and acceptable staleness.
- Your CDN/gateway and existing caching configuration.
- How quickly updates must become visible after a write.