A11y Annotator
Audit HTML for alt text, labels, headings, and ARIA gaps with fix suggestions.
OpenPrice your dependencies in kilobytes before anyone runs npm install.
Select from the curated offline dataset or paste custom sizes.
See approximate min+gzip KB side by side.
Prefer native APIs or lighter libraries where it helps.
Share findings in a PR or architecture note.
Someone on the team suggests adding a charting library and a date helper to a dashboard that already loads slowly. Before anything is installed, it helps to put rough numbers next to each name and see how far over budget the page would end up.
The estimator ships with a built-in table of roughly 64 popular packages, each with an approximate minified size and a minified plus gzipped size. Tick the ones you are weighing up, filter the list by name or category (try "date" or "graphql"), and the report sorts your picks by gzip weight, shows each one's share of the total and compares the sum with a budget you set, 170 KB by default. It also estimates the download time on a slow 4G connection. For private or missing packages, paste lines such as "@acme/ui,42" into the custom box.
Many heavy entries carry a substitution tip, like calling crypto.randomUUID() instead of installing uuid, or picking papaparse over xlsx when you only need CSV. The figures are rounded indicators for a recent major version rather than measurements of your build, so a package that tree-shakes well will usually cost less than listed. To price the imports in an existing file, paste them into the import cost estimator, and browse the web development tools for related helpers.
The package data is part of the page script, so comparisons work offline and your selection is never reported anywhere.
Keyboard shortcut
Answers for this tool. For site-wide help, open the FAQ hub.
The listed size assumes you import the whole package. Bundlers drop unused exports from libraries that support tree shaking, so importing two functions from a modular library costs far less than its full figure.
One package per line written as name,gzipKb, with an optional third value for the minified size. Lines starting with # are ignored. If you leave out the minified size, it is estimated at 3.2 times the gzip figure.
The gzipped total is divided by an effective throughput of about 400 KB per second. Real networks vary and the time spent parsing JavaScript on the device is not included, so use it to compare options rather than to promise a load time.