1.5.3

What's new in v1.5.3

MCP: External link checker

check_external_links pings every external link in a crawl, follows redirects up to 5 hops, and reports dead links (4xx/5xx), redirect destinations, and a summary of OK/redirected/dead counts. Supports domain_filter to scope checks to a specific domain.

MCP: Internal link map

get_link_map returns every crawled page with its internal inlink count. Filter by orphans (0 inlinks) or low (1-3 inlinks) to find pages that are hard to discover. Use show_sources: true to see exactly which pages link to each URL.

MCP: Image auditing

get_images queries all images found during a crawl. Filter by missing_alt to find accessibility issues, or search by src URL, alt text, or page. Results are grouped by page with the alt value shown (or flagged as no alt attribute or empty alt).

MCP: Site structure

get_structure returns the site as a folder tree grouped by URL path segments. Each folder shows page count, average response time, and error count.

MCP: Raw schema data

get_schema returns the raw Schema.org JSON-LD found on pages in a crawl. Filter by URL substring or schema type (e.g. "Product").

MCP: Bulk inlink lookup

get_inlinks now accepts an array of URLs via the urls parameter, so you can check inlinks for multiple pages in a single call instead of firing one request per URL.

MCP: Orphan page detection

get_pages has a new orphan filter that returns indexable pages with zero internal inlinks in one call, replacing the need to check each page individually.

MCP: Full meta description for long meta filter

get_pages with filter: meta_too_long now shows the complete meta description with its character count, so you can read and edit it without a separate lookup.

MCP: Auto-compare crawls

get_issue_diff no longer requires both crawl IDs. Pass only crawl_id_after to automatically compare it against the previous crawl for the same domain, or omit both to compare the last two crawls.

MCP: Ranked linking suggestions

get_linking_suggestions now returns candidates ranked by keyword overlap with suggested anchor text, instead of a raw headings dump. A new target_url mode finds pages that should link TO a given URL, complementing the existing pillar page mode.

Settings: Claude Desktop

Claude Desktop is now available as a one-click option in Settings > AI Tools, alongside Claude Code, Cursor, Codex, and Windsurf.

Bug fixes

• MCP crawl_site now saves external links to the database so check_external_links and the External Links tab both work correctly on MCP-initiated crawls
• MCP schema migrations updated to include all columns added since v1.3, fixing missing data in get_schema, check_external_links, and other tools on fresh installs
• Fixed a bug in get_linking_suggestions where an incorrect column name caused the existing links query to return no results