Changelog

Added

• +Product Data Legibility category (20% weight): measures whether product attributes are rich, granular, and machine-comparable — not just present. Three checks (attribute completeness, granularity, and cross-source consistency) over your JSON-LD and product feed, scoring how legible your catalog is to an AI assistant that shortlists products
• +Discover → Understand → Act framing: scan results now group the seven categories into the arc an agent actually follows — find you, parse your products, then transact
• +Ollama agent backend: the shopping/eval agent can now run on open-weight models (e.g. gpt-oss:120b) via Ollama Cloud, alongside Claude, for cross-model reliability comparisons

Improved

• ↑Scoring rebalanced toward discovery and representation readiness (Discover + Understand = 55%): Crawlability 20%, Schema 15%, Product Data Legibility 20%, Performance 10%, UCP 10%, ACP 10%, MCP 15%. Transaction protocols (UCP/ACP/MCP) are retained as a forward layer rather than the headline
• ↑Landing, results, and docs re-messaged around present-tense AI discovery — "AI assistants are already shortlisting your products"

Added

• +MCP Readiness category (15% weight): checks for Model Context Protocol server discovery endpoints (/.well-known/mcp.json), allowing AI assistants to connect to your store's tools directly
• +6-category weighted scoring: UCP 20% + Schema 20% + Crawlability 20% + Performance 15% + MCP 15% + ACP 10%
• +MCP server: exposes scan_site, get_scan_status, and get_scan_results as MCP tools — AI assistants can trigger and read scans directly
• +Scan artifact collector: persists raw crawl data (HTML, robots.txt, JSON-LD, feeds) to scan_artifacts table for debugging and historical analysis
• +MCP Readiness feature page with protocol overview and check details

Improved

• ↑Category weights rebalanced: UCP, Schema, and Crawlability reduced from 25% to 20% each to accommodate MCP at 15%

Added

• +Multi-page product crawl: scanner now follows up to 10 product page links from your homepage for better Schema.org coverage
• +Redundant check deduplication: multi-page scans now show each URL as a subtitle when the same check runs across multiple pages
• +Plain-English category descriptions in scan results — now explains what each category means for store owners, not just developers
• +E-commerce detection banner: non-commerce sites now receive a clear informational warning in scan results

Fixed

• ~Magento detection false positive: "mage-" substring match was triggering on image-heavy HTML (e.g., university sites). Replaced with specific JS global patterns (var Mage, window.Mage)
• ~USC.edu and similar university sites no longer incorrectly score as e-commerce stores
• ~"AI" and "HTTPS" now display correctly in check names (was "Ai" and "Https")

Added

• +Performance category (15% weight): checks homepage response time, HTTPS enforcement, and redirect chain length
• +Products feed check: validates Shopify /products.json feed accessibility, product data completeness, and variant data
• +5-category weighted scoring: UCP 25% + Schema 25% + Crawlability 25% + Performance 15% + ACP 10%
• +URL reachability pre-check: 422 error returned before scan starts if the target URL is unreachable
• +Frontend URL format validation: invalid URLs are caught before submission

Fixed

• ~Performance scores were blank in scan results due to missing field mapping in API router
• ~Performance category was missing from "All checks" accordion in scan results

Added

• +Solutions nav dropdown with 6 SEO feature pages (UCP Compliance, Monitoring, Competitor Benchmarking, One-Click Fixes, Protocol Tracker, API Access)
• +$19 one-time PDF report tier added to pricing page
• +JWT authentication: register, login, token refresh, /me endpoint
• +Scan history dashboard for logged-in users

Added

• +Public launch: scan any e-commerce URL for free, no signup required
• +UCP compliance checks: manifest existence, structure validation, required fields
• +Schema.org quality checks: JSON-LD presence, Product type, required fields
• +AI Bot Crawlability: robots.txt analysis for 16 AI shopping and search bots
• +ACP readiness signals: Stripe integration, checkout API indicators, catalog feed
• +Overall weighted compliance score (0–100)
• +Results in 15–30 seconds via Celery task queue
• +Railway (API + Worker + PostgreSQL + Redis) + Vercel (frontend) deployment