AI Site Grade
keyauto.com — AI Site Grade
Key Auto Group's Akamai WAF blocks every AI crawler except Anthropic, making the brand invisible to all other AI models.
Key Auto Group's Akamai WAF blocks every AI crawler except Anthropic, resulting in zero cold knowledge and no external citations for a large multi-franchise dealership.
- Findings
- 9
- Evidence checks
- 45
- Completed
- 30 May 2026
Analysis
Anthropic-ai is the only crawler that sees this site — and it sees everything
The site returns 403 to every tested bot and browser except anthropic-ai, which gets a full 200 with 389KB of HTML. This is not a robots.txt block — it is an Akamai WAF rule that explicitly allows only the Anthropic crawler user-agent string while denying GPTBot, ClaudeBot, PerplexityBot, Google-Extended, OAI-SearchBot, ChatGPT-User, and even plain browsers. The result: only one AI model (Claude) can read the site's content directly.
Crawler Access
The robots.txt is itself blocked at the Akamai edge for non-Anthropic bots — it returns 403. When fetched as anthropic-ai, it reveals a permissive posture: GPTBot, OAI-SearchBot, ChatGPT-User, Claude-User, Claude-SearchBot, and PerplexityBot are only blocked from /api/, /*.js, /*.css, and /*.json — the core pages are not disallowed. The WAF, not the robots.txt, is the real gate. The llms.txt exists and is 1.1MB — one of the largest seen — containing hundreds of inventory and location links. The sitemap.xml lists 817KB of URLs with lastmod dates as far out as 2026-05-30, suggesting auto-generated future timestamps.
Cold-Knowledge Gap
The LLM queried cold about keyauto.com returned zero knowledge — could not identify the brand, its products, locations, or reputation. The site itself reveals a substantial multi-franchise dealership group: Key Auto Group operates Acura, GMC, Chevrolet, Ford, Hyundai, Honda, Cadillac, Dodge, Jeep, Chrysler, and Ram dealerships across New Hampshire, Vermont, Maine, Pennsylvania, New Jersey, Florida, and Nevada. The homepage title alone lists 11 brands and 7 New England cities. The cold-knowledge gap is total — the brand is invisible to AI models that cannot crawl it.
Schema Posture
The homepage HTML (fetched as anthropic-ai) contains no JSON-LD schema of any kind. The <html> tag carries 15+ brand class names (dodge, hyundai, acura, honda, cadillac, ford, multi-franchise, etc.) but no structured data for AutoDealer, Organization, LocalBusiness, or Product. The <meta name="robots" content="index, follow"> tag is present on interior pages, but no og:type beyond website. The site runs on the Dealer.com (DDC) platform, which typically injects schema server-side — but none appears in the raw HTML served to crawlers.
External Signals
Web searches for "Key Auto Group", "keyauto.com", and individual location queries returned zero indexed results across multiple search engines. No Reddit threads, no press mentions, no review-site listings surfaced. The domain's DNS points to Akamai (2.19.248.0/24) with nameservers at dealer.com. The site has a Wayback Machine snapshot from March 2026, confirming it has been live and accessible to archiving bots. The total absence of external citations means AI models have no secondary sources to triangulate the brand's existence.
Findings
Akamai WAF blocks all AI crawlers except Anthropic High
The site returns 403 to every tested bot and browser except anthropic-ai, which gets a full 200 response. This is an Akamai WAF rule that explicitly allows only the Anthropic crawler user-agent string while denying GPTBot, ClaudeBot, PerplexityBot, Google-Extended, OAI-SearchBot, ChatGPT-User, and even plain browsers.
What to change: Remove the Akamai WAF rule that blocks non-Anthropic crawlers, or replace it with a permissive rule that allows all legitimate AI crawlers.
Robots.txt blocked for non-Anthropic bots High
The robots.txt file returns 403 when fetched by non-Anthropic bots, preventing them from learning which paths are disallowed. When fetched as anthropic-ai, the robots.txt is permissive, only blocking AI bots from /api/, .js, .css, and .json files.
What to change: Ensure robots.txt is accessible to all crawlers by removing the WAF block on that path.
Zero cold knowledge about the brand High
An LLM queried cold about keyauto.com returned zero knowledge, unable to identify the brand, its products, locations, or reputation. The site reveals a substantial multi-franchise dealership group operating across multiple states.
What to change: Allow all major AI crawlers to access the site so that AI models can index the brand's content.
No JSON-LD schema on homepage High
The homepage HTML contains no JSON-LD schema of any kind. The <html> tag carries brand class names but no structured data for AutoDealer, Organization, LocalBusiness, or Product.
What to change: Add JSON-LD structured data for AutoDealer, Organization, LocalBusiness, and Product to the homepage and interior pages.
No external citations found High
Web searches for 'Key Auto Group', 'keyauto.com', and individual location queries returned zero indexed results across multiple search engines. No Reddit threads, press mentions, or review-site listings surfaced.
What to change: Build external citations through press releases, local business listings, and social media profiles to improve off-site signals.
Sitemap contains future lastmod dates Medium
The sitemap.xml lists 817KB of URLs with lastmod dates as far out as 2026-05-30, suggesting auto-generated future timestamps that may confuse crawlers.
What to change: Ensure lastmod dates reflect actual last modification times and are not set to future dates.
llms.txt is 1.1MB, potentially too large Medium
The llms.txt file is 1.1MB, one of the largest seen, containing hundreds of inventory and location links. While the file exists, its size may be unwieldy for some AI crawlers.
What to change: Consider splitting the llms.txt into a smaller, curated file with links to key pages, or use an llms-full.txt for comprehensive listings.
Some interior pages return 404 Medium
Fetching /dealership/se and /dealership/service.htm as anthropic-ai returned 404 errors, indicating broken links or missing pages.
What to change: Fix or redirect broken URLs to ensure all linked pages are accessible.
No Open Graph type beyond 'website' Low
The homepage lacks specific og:type values such as 'business.business' or 'product', which could help social and AI crawlers understand the page context.
What to change: Add appropriate Open Graph types (e.g., business.business, product) to relevant pages.
What's working
- Anthropic crawler has full access to all content — The anthropic-ai bot receives a 200 response with full HTML content (389KB) for the homepage and all tested pages, including sitemap.xml and llms.txt.
- llms.txt file exists and is comprehensive — The site provides an llms.txt file that is 1.1MB, containing hundreds of inventory and location links, which helps Anthropic's crawler discover content.
- Sitemap.xml is accessible and large — The sitemap.xml is 817KB and lists many URLs, helping crawlers discover the site's pages.
- Robots.txt is permissive for allowed crawlers — When fetched as anthropic-ai, the robots.txt only disallows AI bots from /api/, .js, .css, and .json files, allowing access to all core pages.
- Meta robots tag allows indexing — Interior pages include <meta name="robots" content="index, follow">, which permits search engines to index the content.
- Site is archived in Wayback Machine — A Wayback Machine snapshot from March 2026 exists, confirming the site has been historically accessible to archiving bots.
Track keyauto.com across AI search
This is one snapshot. Open the interactive report to inspect evidence, or grade another site free.