lychee.toml

# Lychee link checker configuration
# https://lychee.cli.rs/configuration/

# Verbose program output
# Accepts log level: "error", "warn", "info", "debug", "trace"
verbose = "error"

# Accept these HTTP status codes as valid
# 403 is included because many sites block automated requests (bot detection)
accept = [200, 204, 301, 302, 307, 308, 403]

# User agent for requests (full Chrome version to avoid bot detection)
user_agent = "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/146.0.0.0 Safari/537.36"

# Don't show interactive progress bar while checking links.
no_progress = false

# Request timeout in seconds
timeout = 30

# Maximum number of retries per request
max_retries = 3

# Wait time between retries (seconds)
retry_wait_time = 10

# Maximum concurrent requests
max_concurrency = 128

# Don't check URL fragments/anchors (e.g. #section-name).
# Fragment validation requires fetching and parsing the full HTML of each target
# page, which greatly increases the number of requests and triggers the known
# lychee SendError panic (race condition during shutdown).
include_fragments = false

# Only test links with the given schemes (e.g. https).
# Omit to check links with any other scheme.
# At the moment, we support http, https, file, and mailto.
scheme = ["https"]

# When links are available using HTTPS, treat HTTP links as errors.
require_https = true

# Exclude patterns (regex)
exclude = [
  # Placeholder/example URLs
  "^https://example\\.com",
  "^https://github\\.com/owner/repo",
  "github\\.com/your-org/",
  # Government sites that block automated requests
  "^https://(www\\.)?fcc\\.gov",

  # Analytics and tracking
  "^https://us\\.i\\.posthog\\.com",
  "^https://c\\.vialoops\\.com",
  "_vercel/(speed-)?insights/",

  # Sites that block automated requests
  "^https://docs\\.stack-auth\\.com",
  "^https://(www\\.)?linkedin\\.com",
  "^https://(www\\.)?twitter\\.com",
  "^https://x\\.com",
  "^https://www\\.npmjs\\.com",
  "^https://cdn\\.simpleicons\\.org",
  "^https://(www\\.)?ctia\\.org",
  "^https://(www\\.)?pcre\\.org",

  # Auto-generated links (Fern "Ask AI" feature)
  "^https://claude\\.ai",

  # GitHub URLs (verified locally or via separate low-concurrency check)
  # Note: This blanket exclude is used when excludeGithub=true in the Node script
  # The script handles GitHub URLs separately to avoid rate limiting
  "^https://github\\.com/",
  "^https://github\\.githubassets\\.com",

  # Static assets (CSS, JS, fonts, images) — no need to link-check these
  "\\.(css|js|woff2?|ttf|eot|svg|png|jpe?g|webp|gif|ico)([?#]|$)",

  # Fern renders a phantom /assets/images/... URL from a CSS comment (fixed in styles.css,
  # but keep this exclude until the fix is deployed and confirmed)
  "/assets/images/\\.\\.\\.",

  # Non-HTTP schemes
  "^mailto:",
  "^tel:",
  "^javascript:",
  "^#"
]

# Browser-like headers to trigger Vercel edge revalidation.
# Without these, Vercel serves stale cache and never revalidates.
# The sec-fetch-* headers signal a real browser navigation vs a bot/crawl.
[header]
accept = "text/html,application/xhtml+xml,application/xml;q=0.9,image/avif,image/webp,image/apng,*/*;q=0.8"
accept-language = "en-US,en;q=0.9"
cache-control = "max-age=0"
sec-fetch-dest = "document"
sec-fetch-mode = "navigate"
sec-fetch-site = "none"
sec-fetch-user = "?1"
upgrade-insecure-requests = "1"
sec-ch-ua = "\"Chromium\";v=\"146\", \"Not-A.Brand\";v=\"24\""
sec-ch-ua-mobile = "?0"
sec-ch-ua-platform = "\"Windows\""