Skip to content

Releases: PO-VINCENT/ai-shopping-audit

Release list

v0.7.0 — calibrated scoring, crawler view modes, checkout-trust rules

v0.7.0 — calibrated scoring, crawler view modes, checkout-trust rules Latest
Latest

Choose a tag to compare

View all tags
@PO-VINCENT PO-VINCENT released this 13 Jul 00:04
v0.7.0
69e49d5

The score now discriminates: every finding deducts by severity (GEO product-data rules weighted highest), hard caps unchanged, floor at 1 — real-world pages that scored 85–95 "ready" now land where they belong (a live supermarket page: 87 → 69 "needs_work").

New since 0.6.0

  • Rendered vs crawler view modes in the extension: two tabs under the score swap the whole popup between what a browsing agent sees and what non-rendering AI crawlers receive — the gap measures your page's JavaScript dependence.
  • Metric taxonomy across all surfaces: eight status tiles (readability, validity, completeness, consistency, trust, accessibility, transactability, freshness) with filtering and grouping — status, never a second score.
  • Agentic-checkout trust rules researched against UCP (ucp.dev), OpenAI's ACP checkout schema, and Microsoft's MMC docs: policy links, seller identity, return-policy and shipping-details completeness.
  • Opt-in --online checks: product-image size minimums (GS1-grade header sniffing, no imaging deps) and IndexNow key verification.
  • 13 more sourced rules including GS1 GTIN validation, ISO 4217, schema.org availability vocabulary, and prompt-injection detection (caps the score at 49).

Full details in CHANGELOG.md · rules with sources in docs/RULES.md

uvx --from catalogready-ai catalogready https://your-store.com/products/example
Assets 3
Loading

v0.6.0 — first public release

v0.6.0 — first public release

Choose a tag to compare

View all tags
@PO-VINCENT PO-VINCENT released this 11 Jul 13:02
v0.6.0
b303222

First public release of CatalogReady — the open-source Lighthouse for AI shopping.

uvx --from catalogready-ai catalogready https://your-store.com/products/example

0–100 readiness score from ~40 deterministic rules (each cited to platform docs), findings with rule IDs, merchant questions instead of invented facts, auto-drafted JSON-LD fixes validated against an in-memory preview, local dashboard + terminal chat + Chrome extension + MCP server, English/中文 UI. Offline, no API key, never writes to your store.

Full details in CHANGELOG.md. Quick starts for ChatGPT/Claude/Gemini/Copilot/DeepSeek in docs/QUICKSTART-AI-ASSISTANTS.md.

Loading