AI logo
AI

rspeedy-bundle-size

Helps coding agents analyze and reduce the shipped .lynx.bundle size of Rspeedy and ReactLynx apps. It is a measure-first skill: agents should identify where the bytes live before recommending or applying optimizations.

Anatomy of a Lynx dual-thread bundle

Use it when an agent needs to:

  • explain why a Rspeedy or ReactLynx bundle is large
  • break bundle size down across media assets, background-thread JavaScript, and main-thread JavaScript
  • inspect rsdoctor or stats.json output before choosing an optimization
  • reduce main-thread leakage by moving background-only code out of the render path
  • evaluate size levers such as media compression, duplicate-package cleanup, extractStr, lazy bundles, or dynamic component splitting

Installation

npx skills add lynx-community/skills -s rspeedy-bundle-size

This installs the rspeedy-bundle-size skill so compatible coding agents can load its measurement workflow, gotchas, and bundled analysis helpers.

What It Covers

  • Measurement discipline — using Rspeedy stats or rsdoctor data before proposing fixes.
  • Layer-aware analysis — separating media assets, react:background, and react:main-thread weight.
  • Importer tracing — using real module graph edges instead of guessing why a module is bundled.
  • Optimization levers — prioritizing assets, background JS, main-thread leakage, and compile-layer knobs by expected impact.
  • Bundle optimization logs — recording before/after .lynx.bundle size, build commands, caveats, and shipped changes.

Ask the agent to use rspeedy-bundle-size before changing code for a bundle-size request.

  1. Read the repository build setup and use the native build command, such as Rush, wrapper tooling, or a package script.
  2. Generate a real size breakdown with rsdoctor or Rspeedy stats.
  3. Report the largest lever first, with evidence and tradeoffs.
  4. Change app code only when requested, and keep every applied optimization backed by a reproducible before/after measurement.

Learn More

Except as otherwise noted, this work is licensed under a Creative Commons Attribution 4.0 International License, and code samples are licensed under the Apache License 2.0.