lychee.toml

# Increase timeout for AMD PDF and documentation links, which can be slow
timeout = 30

# Retry transient failures (e.g., GitHub rate limiting, flaky CDN)
max_retries = 3

# Accept 429 (rate limited) as non-failure — the link exists, we're just throttled
# Accept 206 (partial content) for PDF/media links
accept = [200, 206, 429]

# Skip anchor fragment validation — anchors change frequently and cause false positives.
# lychee's --include-fragments is now a string enum (none|anchor-only|text-only|full),
# not a bool — `false` fails to parse on current lychee releases.
include_fragments = "none"

# Exclude patterns that reliably cause false positives with automated checkers
exclude = [
  # LinkedIn blocks automated link checks
  "^https?://(www\\.)?linkedin\\.com",
  # Reddit returns 403 to bots/CI environments
  "^https?://(www\\.)?reddit\\.com",
  # AMD's PDF CDN is slow and often returns transient errors to bots
  "^https?://www\\.amd\\.com/content/dam/.*\\.pdf",
  # AMD's account/registration host rejects lychee's HTTP/2 requests
  "^https?://account\\.amd\\.com/",
  # Local/relative anchors
  "^#",
  # Raw HTML nav links in the instinct-design cards point at post-build
  # `.html` paths (e.g. system-admin/co-tools.html) that only exist in the
  # Sphinx build output, not in the markdown source tree — can't be
  # validated without a full docs build, so skip local .html targets.
  "^file://.+\\.html(#.*)?$",
]

# Do not check generated build output or version control metadata
exclude_path = [".git", "docs/_build"]

# User agent to use for requests
user_agent = "lychee"