Recker is a batteries-included HTTP client for Node.js that eliminates the need to juggle multiple libraries. Unlike minimal clients like ky or got that require bolting on separate packages for retries, caching, rate limiting, and authentication, Recker ships with 30+ built-in plugins covering everything from OAuth2 and circuit breakers to HLS streaming and gRPC-Web. It provides a unified, fluent API for HTTP/1.1, HTTP/2, HTTP/3, WebSockets, SSE, and even FTP/SFTP—so you don't need different clients for different protocols. Performance is a first-class concern: Recker uses connection pooling, automatic keep-alive, and smart concurrency controls out of the box. For testing, it includes comprehensive mock servers for every protocol it supports, plus a full MCP server for AI-powered debugging. Whether you're building a simple REST client or a complex multi-protocol integration, Recker gives you production-ready defaults with zero configuration while remaining fully customizable when you need it.
Report
@filipeforattini Love the batteries‑included approach. How do you see Recker competing with minimal clients in performance benchmarks?
Report
Maker
Hey @masump — great question! Back when I started building Recker, the goal was simple: get undici-level performance without forcing developers to assemble a dozen plugins just to reach production quality. In 2026, with systems becoming more distributed, API budgets shrinking, and multi-protocol integrations becoming the norm, that goal matters even more.
The good news: Recker holds up extremely well against minimal clients. In the benchmarks I published (https://forattini-dev.github.io/recker/#/benchmarks), Recker is faster than axios, got, ky and others, and stays very close to undici’s raw engine — even with 30+ built-in features.
What really makes a difference today is what minimal clients don’t cover: dedup, caching, client pooling, backpressure, retries with real jitter, cross-protocol consistency. In modern workloads — parallel traffic, multi-domain fan-out, duplicate request storms — these features make Recker 2–3× more efficient than minimal clients.
@filipeforattini Love the batteries‑included approach. How do you see Recker competing with minimal clients in performance benchmarks?
Hey @masump — great question! Back when I started building Recker, the goal was simple: get undici-level performance without forcing developers to assemble a dozen plugins just to reach production quality. In 2026, with systems becoming more distributed, API budgets shrinking, and multi-protocol integrations becoming the norm, that goal matters even more.
The good news: Recker holds up extremely well against minimal clients. In the benchmarks I published (https://forattini-dev.github.io/recker/#/benchmarks), Recker is faster than axios, got, ky and others, and stays very close to undici’s raw engine — even with 30+ built-in features.
What really makes a difference today is what minimal clients don’t cover: dedup, caching, client pooling, backpressure, retries with real jitter, cross-protocol consistency. In modern workloads — parallel traffic, multi-domain fan-out, duplicate request storms — these features make Recker 2–3× more efficient than minimal clients.