Scanner

    Configuration

    Every call takes an optional second argument. Anything you leave out falls back to the exported DEFAULTS.

    import { audit, DEFAULTS } from "safi-studio-scanner";

console.log(DEFAULTS);

const report = await audit("https://example.com", {
  maxPages: 50,
  concurrency: 8,
  browser: true,
});

    Options

    OptionTypeDefaultWhat it does
    maxPagesnumber20Hard cap on pages crawled. The crawl stops at this count even if more are linked.
    concurrencynumber5Pages fetched at the same time. Also caps the link checker's request pool.
    maxDepthnumber3How many link hops from the start URL to follow. Depth 0 is the start page.
    onlystring[]allCategories to run. Everything else is skipped.
    skipstring[]noneCategories to leave out. Applied after only.
    timeoutnumber15000Per-request timeout in milliseconds, for pages and link checks alike.
    userAgentstringSafiStudioScanner/0.1 (+https://safi-studio.com)Sent on every request. Change it if a host blocks the default.
    browserbooleanfalseRender each page in local Chromium to collect axe-core results and Core Web Vitals.
    psiKeystringunsetUse the PageSpeed Insights API instead of a local browser. Ignored when browser is true.
    psiMaxPagesnumber5How many pages to send to PageSpeed Insights. The API is rate limited, so this stays low.
    format"json" | "md" | "html""html"Only read by render() and the auditTo* helpers. audit() always returns the object.

    Tuning the crawl

    One page. { maxPages: 1 }. The fastest useful run, and the right choice when you audit a specific URL rather than a site.

    A section. Point the start URL at the section and keep maxDepth low. The crawler stays on the same origin, so a link to another domain is checked for reachability but never crawled.

    A whole small site. Raise maxPages to the page count you expect and maxDepth to 5 or more. Watch report.pagesScanned to confirm the crawl reached everything.

    A large site. Raise concurrency before you raise maxPages. Be careful: the link rules also fetch every unique outbound URL once, so a page with 200 links generates 200 requests regardless of the crawl settings.

    Be polite

    Concurrency applies to a single host. A high value on a small server is a load test nobody asked for. Start at 5 and raise it only against infrastructure you own or have permission to hit.

    Choosing categories

    Categories map to rule id prefixes:

    core-seo · content · links · images · structured-data · security · crawlability · url-structure · social-media · internationalization · legal · analytics · eeat · performance · accessibility

    // A fast pass with no extra network requests: skip the two that fetch other URLs.
await audit(url, { skip: ["links", "images"] });

// A security review only.
await audit(url, { only: ["security", "crawlability"] });

    only runs first, then skip removes from what is left. Passing both is allowed but rarely what you want.

    Defaults, verbatim

    export const DEFAULTS: AuditOptions = {
  format: "html",
  timeout: 15000,
  userAgent: "SafiStudioScanner/0.1 (+https://safi-studio.com)",
  maxPages: 20,
  concurrency: 5,
  maxDepth: 3,
  browser: false,
  psiMaxPages: 5,
};