Scovant Core — https://example.com/step4

Profile: commerce (requested auto, confidence 85%)

Core version: 0.1.0

1 PASS, 23 WARN, 8 FAIL, 13 N/A, 0 ERROR (45 checks)

Score

35 / 100

Grade: F · Coverage: 100%

Categories

Findings

Experimental (not scored)

• CORE-INTERFACE-008 (WARN) — A UCP profile is published but fails validation.

N/A: CORE-INTERFACE-004, CORE-INTERFACE-009, CORE-MACHINE-012, CORE-OPERABILITY-007

Evidence

{
  "final_url": "https://example.com/step4",
  "input_url": "https://example.com/",
  "redirect_chain": [
    "https://example.com/",
    "https://example.com/step1",
    "https://example.com/step2",
    "https://example.com/step3"
  ],
  "redirect_count": 4,
  "status": 200
}

{
  "error": null,
  "resource": "https://example.com/robots.txt",
  "served_as_html": false,
  "sha256": "901c28336f0ee2ceb8b00545721bc3a75c09d539b54d031ab78f34d43f7e5294",
  "sitemap_count": 0,
  "status": 200,
  "unknown_directives": []
}

{
  "declared_policy": {
    "Bingbot": false,
    "Claude-SearchBot": false,
    "Googlebot": false,
    "OAI-SearchBot": false,
    "PerplexityBot": false
  },
  "http_status": 200,
  "resource": "https://example.com/robots.txt",
  "robots_present": true
}

{
  "explicit_separation": false,
  "http_status": 200,
  "resource": "https://example.com/robots.txt",
  "search_blocked": [
    "OAI-SearchBot",
    "Claude-SearchBot",
    "PerplexityBot",
    "Googlebot",
    "Bingbot"
  ],
  "training_blocked": [
    "GPTBot",
    "ClaudeBot",
    "Google-Extended",
    "CCBot",
    "Applebot-Extended"
  ]
}

{
  "entry_count": 0,
  "exists": false,
  "kind": null,
  "parse_error": null,
  "probe_error": null,
  "probe_status": 404,
  "served_as_html": false,
  "url": null,
  "valid": false
}

{
  "url": null
}

{
  "canonical_url": "https://example.org/elsewhere",
  "entry_url": "https://example.com/step4"
}

{
  "robots_meta": "noindex",
  "x_robots_tag": null
}

{
  "errors": [],
  "references_broken": [
    "https://example.com/does-not-exist"
  ],
  "references_checked": 1,
  "references_unresolved": [],
  "resource": "https://example.com/llms.txt",
  "valid": true
}

{
  "declared": true,
  "dimensions": {
    "ai-input": "unset",
    "ai-train": "unset",
    "search": "unset"
  },
  "syntax_errors": [
    "unrecognized value for 'search': 'maybe' (expected yes|no)"
  ]
}

{
  "exists": false,
  "http_status": 404,
  "resource": "https://example.com/.well-known/mcp.json",
  "server_card": false,
  "valid": false
}

{
  "pages_scanned": 1,
  "pages_with_marker": []
}

{
  "pages": [
    {
      "parsed": 1,
      "raw": 2,
      "url": "https://example.com/step4"
    }
  ],
  "parsed_total": 1,
  "raw_total": 2
}

{
  "found": false,
  "pages_parsed": 1
}

{
  "found": false,
  "pages_parsed": 1
}

{
  "has_identifiers": false,
  "pages_parsed": 1,
  "product_pages": [
    "https://example.com/step4"
  ]
}

{
  "availability": null,
  "currency": null,
  "missing": [
    "price",
    "priceCurrency",
    "availability"
  ],
  "price": null,
  "url": "https://example.com/step4"
}

{
  "count": 0,
  "identifier_keys": [],
  "pages_parsed": 1
}

{
  "entry_url": "https://example.com/step4",
  "issues": [
    "og:title",
    "og:description"
  ],
  "meta_description": "A synthetic fixture site with several broken agent-readiness signals.",
  "missing": [
    "og:title",
    "og:description"
  ],
  "title": "Example Bad Shop"
}

{
  "entry_url": "https://example.com/step4",
  "h1_count": 2,
  "heading_count": 2,
  "issues": [
    "multiple H1"
  ],
  "levels": [
    "h1",
    "h1"
  ]
}

{
  "entry_url": "https://example.com/step4",
  "html_lang": null
}

{
  "covered": 0,
  "empty_alt": 0,
  "pages_parsed": 1,
  "ratio": 0.0,
  "total": 1,
  "with_alt": 0
}

{
  "entry_url": "https://example.com/step4",
  "spa_shell_marker": false,
  "visible_text_chars": 148
}

{
  "redirect_chain": [
    "https://example.com/",
    "https://example.com/step1",
    "https://example.com/step2",
    "https://example.com/step3"
  ],
  "redirect_count": 4
}

{
  "cache_control": null,
  "etag": null,
  "last_modified": null
}

{
  "broken": [
    {
      "source": "canonical",
      "status": 404,
      "url": "https://example.org/elsewhere"
    },
    {
      "source": "llms",
      "status": 404,
      "url": "https://example.com/does-not-exist"
    }
  ],
  "broken_count": 2,
  "inconclusive": [],
  "refs_checked": 2,
  "refs_resolved": 2,
  "unresolved": []
}

{
  "dom_nodes": 16,
  "estimated_tokens": 37,
  "html_bytes": 725,
  "level": "LOW",
  "link_count": 0,
  "script_bytes": 88,
  "script_ratio": 0.12137931034482759,
  "structured_bytes": 88,
  "text_chars": 148,
  "token_chars_ratio": 4
}

{
  "totals": {
    "forms": 1,
    "inputs": 1,
    "unlabeled_inputs": 1,
    "unlabeled_selects": 0,
    "unnamed_buttons": 0
  }
}

{
  "contact_url": null,
  "kind": null
}

{
  "contact": false,
  "expires": "2020-01-01T00:00:00Z",
  "expires_valid": false,
  "found_url": "https://example.com/.well-known/security.txt",
  "status": 200
}

Remediation

• CORE-ACCESS-001 — Collapse the redirect chain to at most one hop.
• CORE-ACCESS-002 — Scope Disallow rules to specific paths instead of blocking `/` for `*`.
• CORE-ACCESS-003 — Allow search/retrieval crawlers (e.g. OAI-SearchBot, Claude-SearchBot, PerplexityBot) in robots.txt while keeping any training restrictions separate.
• CORE-ACCESS-004 — Restrict only the training-purpose crawler tokens; leave search/retrieval crawlers allowed.
• CORE-ACCESS-005 — Publish /sitemap.xml and declare it in robots.txt.
• CORE-ACCESS-007 — Point the canonical tag at a URL on this site's own host.
• CORE-ACCESS-008 — Remove the noindex directive from the meta robots tag or X-Robots-Tag header.
• CORE-ACCESS-009 — Remove or fix the broken references in llms.txt.
• CORE-ACCESS-010 — Fix the Content-Signal directive, e.g. "Content-Signal: search=yes, ai-input=yes, ai-train=no".
• CORE-MACHINE-001 — Fix the malformed JSON-LD blocks so every one parses (validate with a JSON linter before publishing).
• CORE-MACHINE-002 — Add JSON-LD with "@type": "Organization" declaring at least name and url.
• CORE-MACHINE-003 — Add JSON-LD with "@type": "WebSite" or "WebPage".
• CORE-MACHINE-004 — Add a stable identifier (sku, gtin, mpn, or brand) to the Product entity.
• CORE-MACHINE-005 — Add an Offer with price, priceCurrency, and availability to the Product entity.
• CORE-MACHINE-006 — Add a stable identifier (sku, gtin, mpn, or brand) to the Product entity.
• CORE-MACHINE-008 — Add the missing metadata tags and give title and description distinct content.
• CORE-MACHINE-009 — Use exactly one H1 per page and avoid skipping heading levels (e.g. h1 to h3).
• CORE-MACHINE-010 — Add a valid `lang` attribute to the `<html>` tag (e.g. `lang="en"`).
• CORE-MACHINE-011 — Add an alt attribute (or an explicit empty alt for decorative images) to every image.
• CORE-OPERABILITY-001 — Server-render (or statically pre-render) the core content so an agent's plain HTTP fetch sees it.
• CORE-OPERABILITY-002 — Collapse the redirect chain to at most one hop.
• CORE-OPERABILITY-003 — Add an ETag or Last-Modified header so agents can issue conditional GETs.
• CORE-OPERABILITY-004 — Fix or remove the broken references (sitemap, llms.txt, OpenAPI spec, policy links) so agents don't hit dead links.
• CORE-OPERABILITY-006 — Add a <label> (or aria-label/aria-labelledby) to every input/select, and a name or accessible text to every button.
• CORE-TRUST-001 — Add a clearly labeled contact or support link (a page or a mailto link) to the entry page.
• CORE-TRUST-002 — Publish a shipping policy page (cost, timing, carriers) and link it from the entry page's nav or footer.
• CORE-TRUST-003 — Publish a returns/refund policy page and link it from the entry page's nav or footer.
• CORE-TRUST-004 — Publish a privacy policy page and link it from the entry page's nav or footer.
• CORE-TRUST-005 — Publish a terms/conditions page and link it from the entry page's nav or footer.
• CORE-TRUST-006 — Add a Contact field (RFC 9116) to security.txt.

Limitations

• Scovant Core evaluates declared and static evidence only; it does not observe real agent traffic.
• Only reachability and syntax are checked; per-agent policy is evaluated by the other CORE-ACCESS checks.
• Only the declared robots.txt policy is evaluated; whether the crawler is actually served is not observed.
• Only the declared robots.txt policy is evaluated.
• Only the first-declared sitemap (or its first child, for a sitemap index) is fetched and validated.
• Only ISO 8601 date-formatted lastmod values are parsed; malformed dates are ignored, not penalised.
• Only the entry page's declared canonical is evaluated.
• Only the entry page's own robots meta tag and X-Robots-Tag header are checked.
• Absence is not penalised; the convention is emerging. Reference checks are HTTP status only.
• Absence is not penalised; the convention is emerging.
• Only `/.well-known/mcp.json` and the two candidate server-card paths are probed; a custom discovery location is not found.
• Only the `mcpServers` object at /.well-known/mcp.json is parsed; a server card is not covered by this check.
• This is a static text scan of already-fetched HTML for a marker string; the browser-side registration was not executed, so a marker's presence does not confirm the tools actually register or work.
• Only a fixed set of conventional paths, plus same-origin entry-page links naming openapi/swagger, are probed.
• Only the conventional /.well-known/oauth-authorization-server path is probed.
• Only the conventional /.well-known/oauth-protected-resource path is probed.
• Only well-formed-JSON parseability is checked; schema.org vocabulary correctness is not validated.
• Only the schema_org nodes on the sampled pages are checked; an Organization declared elsewhere on the site is not evaluated.
• Only the schema_org nodes on the sampled pages are checked.
• Only the sampled pages are checked; a product catalog not represented in the sample is not evaluated.
• Only the first sampled page that exposes a Product entity is checked.
• Only the non-entry pages in the sampled set are checked.
• Only the entry page's metadata is checked.
• Only the entry page's heading outline is checked.
• Only the entry page's `<html lang>` attribute is checked.
• Only the images on the sampled pages are counted.
• Static-HTML signal only: the raw fetched response is inspected, never rendered in a browser, so a site that hydrates real content very quickly may still be flagged here.
• Only the entry URL's own redirect chain is inspected; redirects encountered while fetching other sampled pages are not counted here.
• Only the entry response's own headers are checked; per-page-type validator strategy is not inspected.
• Only the capped set of references collected by the machine_links gatherer are checked; endpoints not linked from any declared document are not found. A declared MCP endpoint is recorded but never judged — Core does not perform the MCP handshake.
• `estimated_tokens` is a character-count estimate (chars / token_chars_ratio), never a real tokenizer count, and only the entry page is measured.
• Only forms on the sampled page set are inspected; a form behind client-side rendering that never appears in the static HTML is not seen.
• Only the entry page's anchors are scanned; a contact method reachable only from a deeper page is not found.
• Only a same-origin link discovered from the entry page's nav/footer/anchors is followed; a shipping policy reachable only from a deeper page is not found.
• Only a same-origin link discovered from the entry page's nav/footer/anchors is followed; a returns policy reachable only from a deeper page is not found.
• Only a same-origin link discovered from the entry page's nav/footer/anchors is followed; a privacy policy reachable only from a deeper page is not found.
• Only a same-origin link discovered from the entry page's nav/footer/anchors is followed; terms reachable only from a deeper page are not found.
• Only the conventional /.well-known/security.txt and legacy /security.txt paths are probed.
• Pricing-page discovery follows only a same-origin link from the entry page; the structured-price fallback checks only the sampled pages.

Not tested by Scovant Core

• Observed WAF access
• Real agent tasks
• MCP tool execution
• WebMCP state parity
• Multi-model reliability
• Regression stability

Core vs Cloud

Core measures what a site declares. Cloud measures what real agents actually experience.

Methodology

Category weights are 25/25/20/15/15. Each check contributes PASS = 1, WARN = 0.5, FAIL = 0 to its category; ERROR lowers coverage (a document that could not be read counts against evidence), and N/A is excluded entirely. A category scores INSUFFICIENT_EVIDENCE below a 60% coverage floor of its applicable weight. Full reference: docs/methodology.md.

Provenance

scan_id

local-golden

started_at

2026-09-04T00:00:00Z

completed_at

2026-09-04T00:00:00Z

schema_version

1.0

core_version

0.1.0

ruleset_version

2026.09

ruleset_digest

d83df799eaa7

python

<runtime>

platform

<runtime>

user_agent

ScovantCore/0.1.0 (+https://github.com/Scovant/scovant-core)

timeout

60.0

max_pages

5

network_mode

fixture

experimental

False

allow_private_networks

False