Changelog

What shipped, newest first. RankForge is built in public by a solo founder.

  1. Monitoring, regression gates and a crawler that reads sites like Google

    • Monitoring: RankForge checks your production site on a schedule and alerts you by email, Slack, Discord or webhook when a page starts failing, turns noindex, gets blocked by robots.txt, or when the sitemap or the HTTPS certificate breaks.
    • rankforge diff and watch compare two versions of a site, page by page, and exit 1 on critical changes. The GitHub Action takes a baseline-url to gate pull requests: preview against production.
    • Crawler: sitemap-seeded, up to 500 pages, robots.txt read the way Googlebot reads it, X-Robots-Tag headers, hreflang, canonical targets, soft 404s, sitemap hygiene and certificate checks.
    • Core Web Vitals of real users from Google's Chrome UX Report, and 21 AI crawlers checked for AI-search access.
    • A Claude Code plugin with the MCP tools and two skills, and a compare_sites tool for your editor.
    • The GitHub Action comments the result on the pull request, in one comment updated on every push.

  2. Open-source CLI + free audit page

    • npx rankforge-cli audit <url>: the deterministic engine as a zero-dependency terminal CLI, MIT licensed, with a GitHub Action for CI gates (--fail-under).
    • A free, no-signup audit at /audit: 3 pages, every issue measured with evidence, nothing stored.
    • Engine: crawled pages are deduplicated by final URL so a redirect target is never reported twice.

  3. The RankForge agent in your editor (MCP)

    • RankForge now ships as an MCP server at /api/mcp. Connect it to Claude Code, Cursor, VS Code or any MCP client.
    • Your coding assistant runs the deterministic audit against localhost or production, gets fix templates, and patches the repo it already has open.
    • Deterministic audit engine: measured image weights, real link statuses, internal link graph and content analysis, with no LLM required.

  4. Site-wide review pass

    • Docs pages for every audit rule, framework guides, and this changelog.
    • Security hardening: stricter headers, safer redirects, server-side auth checks.
    • Performance: lighter middleware, faster first paint on the landing page.

  5. Interactive dot systems

    • Living crawler swarm, dotted flow-field currents and a symptom→cause web across the landing page.
    • Full prefers-reduced-motion support with settled static compositions.

  6. Agent swarms + brain

    • In-app audit and fix swarms powered by the Anthropic API.
    • Self-contained semantic memory (local embeddings + Supabase) that grounds audits, fixes and chat.

  7. Supabase accounts

    • Email + password auth, per-user data with row-level security, settings persistence.