Configuration reference

sitecheck reads sitecheck.toml from --config, else from the current directory, else from the audited directory. Unknown sections, unknown keys, wrong types and unknown rule names are rejected with exit code 2. Command-line flags override the file.

[crawl]
max_pages = 500          # pages parsed; further pages are still fetched so links to them verify, but not parsed
max_depth = 10           # link-following depth from the start page (sitemap URLs and files on disk start at 0)
concurrency = 8          # simultaneous HTTP requests (filesystem mode is sequential)
timeout = 15.0           # seconds per HTTP request
retries = 2              # extra attempts on connection errors, 429, 502, 503, 504
backoff = 0.25           # first retry waits backoff seconds, then doubles
respect_robots = true    # HTTP mode: do not crawl pages disallowed for the user agent
user_agent = "sitecheck/0.1"
site_url = ""            # public origin, e.g. "https://example.com"
exclude = []             # glob patterns matched against URL paths, e.g. ["/drafts/*"]

[external]
enabled = false          # same as --external
concurrency = 8
rate_per_host = 2.0      # requests per second per host
timeout = 10.0
ignore = []              # glob patterns matched against full URLs

[budgets]
max_page_weight_kb = 1024
max_requests = 50
max_asset_kb = 500       # any single asset
gzip_min_kb = 20         # minimum size before text-uncompressed is considered
lazy_after = 3           # images at index >= this should be lazy
oversize_factor = 2.0

[seo]
title_min = 10
title_max = 60
description_min = 50
description_max = 160

[severity]               # error | warning | info | off
broken-anchor = "warning"
og-incomplete = "off"

[ignore]
paths = ["/404.html"]    # glob patterns on the page path; issues on matching pages are dropped
codes = []               # rule codes to drop entirely

site_url

Static builds usually contain absolute URLs on the production domain (canonical, og:url, sitemap entries). site_url tells sitecheck that this origin is "the site", so those URLs are verified against the files or server being audited instead of being treated as external. If you do not set it:

• filesystem mode adopts the origin that most sitemap.xml entries use, otherwise http://localhost;
• HTTP mode treats the sitemap's origin as an alias of the audited origin.

An https:// site origin enables the mixed-content checks.

Command line

sitecheck check TARGET [options]
  --config FILE            --external / --no-external   --site-url URL
  --max-pages N            --max-depth N                --concurrency N
  --no-robots              --only links,seo,a11y,perf   --quiet
  --fail-on error|warning|info|none      (default: error)
  --baseline FILE          --write-baseline FILE
  --json FILE  --junit FILE  --sarif FILE  --html FILE
sitecheck rules [--category links|seo|a11y|perf]

Exit codes: 0 nothing at or above --fail-on; 1 at least one issue at or above it; 2 usage, configuration or crawl-start errors (missing directory, unreachable start URL).

Baselines

--write-baseline FILE records the fingerprint code|page|target of every current issue and exits 0. --baseline FILE later hides issues with a recorded fingerprint, so CI fails only on new problems. Fingerprints ignore line numbers and message text, so editing a page does not resurrect an old issue; moving an issue to another page or target does.