sitevet-cli 0.3.1

Command-line website audits built on the sitevet library.
sitevet-cli-0.3.1 is not a library.

sitevet

A website audit engine in Rust. Point it at a URL and it crawls the site, checks every page against 94+ rules for SEO, security, accessibility and performance. You get back a scored report as HTML, Markdown or JSON.

It ships two ways from one codebase:

  • A library crate. Add it to your Rust project, call audit(), and get a typed report back. Stream progress events into your own UI, add your own rules, render the report yourself.
  • A CLI. sitevet audit https://example.com for a quick check in the terminal, or with --fail-under 85 as a CI step that fails when the score drops.

[!NOTE] sitevet 0.3.1 is out. More features are on the way, and build progress is in PROGRESS.md.

Install

With Rust installed:

cargo install sitevet-cli

Or a prebuilt binary, on macOS and Linux:

curl --proto '=https' --tlsv1.2 -LsSf https://github.com/Abdulkader-Safi/sitevet/releases/latest/download/sitevet-cli-installer.sh | sh

Windows (PowerShell):

irm https://github.com/Abdulkader-Safi/sitevet/releases/latest/download/sitevet-cli-installer.ps1 | iex

Binaries for macOS (arm64, x86_64), Linux (arm64, x86_64, static) and Windows (x86_64) are on the releases page. To use the library in your own project: cargo add sitevet (add --features browser for the rendered checks).

What works today

The library and the sitevet command:

  • A breadth-first crawl of same-origin pages with page and depth limits, per-host concurrency, an optional per-host delay, manual redirects and size caps.
  • robots.txt obeyed by default, with blocked URLs listed in the report. Sitemaps (and sitemap indexes) seed the crawl.
  • A shared link checker that requests each URL once per audit.
  • Private-network blocking when the start URL is public.
  • Three entry points: audit(), audit_stream() with typed progress events, and spawn() for apps that don't run tokio.
  • The 94 rules ported from the TypeScript package, with the same IDs, severities and scoring.
  • Rendered checks with --browser (your installed Chrome or Chromium): an axe-core accessibility pass and Core Web Vitals per page. Without Chrome, a PageSpeed Insights key (SITEVET_PSI_KEY) fills the same checks for the first 5 pages. With neither, those 7 rules are listed as skipped and don't touch the score.
  • --render-crawl to find links that only exist after JavaScript runs.
  • Per-page measurements (title and description lengths, link and heading tables, Open Graph and Twitter tags, directives, dates) that never touch the score.
  • Word counts and readability per page, with keyword tables, in the page's language. The language comes from <html lang>, or is detected from the text when that is missing. Stop words and word stems cover every language the ISO stop word lists and the Snowball stemmers do (English and Arabic are tested). English pages get Flesch reading ease, Flesch-Kincaid and Gunning Fog. Other languages get average sentence and word length instead, and pages in Chinese, Japanese, Thai or Khmer get a character count.
  • Answer engine measurements per page: a definition or short answer near the top, question headings and how many get a direct answer, answer formats (lists, steps, tables, definitions, FAQ blocks), words per H2 section, internal links in the main content, AI crawler access from robots.txt, llms.txt, how much content is there without JavaScript, and a page-type hint. They never touch the score.
  • A structured data tree read from JSON-LD, microdata and RDFa alike, with the key properties each item lacks, plus FAQPage and HowTo checks.
  • Your own rules through the Rule trait. See CONTRIBUTING.md and examples/custom-rule.rs.
  • Reports as JSON (with a JSON Schema), Markdown, or one self-contained HTML file with light and dark themes.
  • The sitevet CLI: audit with a flag for every option, sitevet.toml, progress on stderr, --fail-under and --fail-on for CI, plus rules and init.
  • Run history: every CLI audit is saved locally. sitevet history lists runs, sitevet show <id> renders one again, sitevet diff <a> <b> compares two, and --baseline last adds a "changes since" section to the report. --no-save skips saving.

Planned: everything listed below.

Where it comes from

sitevet is a Rust rebuild of an earlier TypeScript audit package by the same author, which runs 94 rules across 15 categories. That package stays maintained in its own repo. sitevet starts from the same rule catalog and rule IDs, then adds most of that project's roadmap in its first release.

What v1 checks

The 94 rules from the TypeScript package, covering core SEO, content, links, images, structured data, security headers, crawlability, URL structure, social tags, hreflang, legal pages, analytics, E-E-A-T signals and performance. With a local Chrome install (or a PageSpeed Insights key), it adds a full axe-core accessibility pass and Core Web Vitals.

On top of that, v1 plans to ship:

  • Site-wide findings: duplicate titles and descriptions, near-duplicate pages, orphan pages, click depth, an internal link graph.
  • Trust signals and local SEO, such as author bylines, refund pages, and NAP consistency with LocalBusiness schema.
  • Tech stack detection for analytics vendors, ad pixels, the CMS and the CDN.
  • Deeper performance timings, mobile checks (tap targets, overflow, console errors), screenshots and an opt-in image size audit.

Every page gets two kinds of output. Findings are verdicts (pass, fail, warn) with the offending markup as evidence, and they move the score. Measurements are numbers and tables, like title length or the full heading outline, and they never touch the score.

Usage

As a library (works today; enable the browser feature for rendered checks):

let report = sitevet::audit("https://example.com", sitevet::Options::default()).await?;
println!("{} pages, score {}", report.pages.len(), report.score);

From the terminal:

sitevet audit https://example.com --out report.html
sitevet audit https://example.com --format json --fail-under 85 > report.json
sitevet audit https://example.com --browser --out report.html   # needs Chrome or Chromium
sitevet rules --category security
sitevet history                                   # saved runs, newest first
sitevet audit https://example.com --baseline last --out report.html
sitevet init

The report goes to stdout, or to --out (the format follows its extension unless --format says otherwise). Progress goes to stderr. The exit code is 0 when the audit ran and passed, 1 when --fail-under or --fail-on tripped, and 2 when it could not run.

Settings can live in a sitevet.toml in the current folder (sitevet init writes one with every default). Its keys match the library options one to one, and command-line flags override it. The full rule list is in docs/rules.md. History lives in history.db under SITEVET_HOME, or the OS data folder (~/Library/Application Support/sitevet, ~/.local/share/sitevet, %APPDATA%\sitevet).

License

MIT. Copyright Abdulkader Safi.