Cache and accelerate content delivery with intelligent purging, on-the-fly compression, image optimization, and edge computing capabilities. Written in Koder Lang.
Memory (hot) and disk (warm) storage tiers with LRU, LFU, and ARC eviction policies. Automatic promotion from disk to memory on access.
Purge by URL, wildcard pattern, cache tag, or regex ban. Soft purge marks content stale without removing it. Batch purge for bulk operations.
Transparent gzip, Brotli, and Zstandard compression with Accept-Encoding negotiation. Compressible content types auto-detected.
Resize, quality adjustment, and automatic WebP/AVIF conversion. Query parameter API for on-demand transformations.
Run Koder Lang snippets at the edge with sandboxed execution. Transform requests and responses with custom logic.
Varnish-inspired configuration with recv, hash, hit, miss, deliver, and synth phases. Full request/response control.
Hit ratio, bandwidth savings, latency percentiles (P50/P95/P99), top URLs, and status code distribution with Prometheus export.
Serve stale content when origin is down. Supports stale-while-revalidate and stale-if-error Cache-Control directives.
Deduplicate concurrent requests for the same resource. Only one request goes to origin; all others wait for and share the result.