Verification Domains

14 domains. One API. Financial is the validated domain — 91.47% match-rate (n=938), computed 2026-10-04, published at /v1/accuracy/financial; the other 13 are not-yet-validated. 6 of the 14 domains return a verdict, confidence score, and evidence: brand, competitive, compliance, financial, policy, spec.

How this count is derived. 14 is the number of distinct {domain} names among the POST /v1/{domain}/check paths in the API's own OpenAPI document (rule ^/v1/([a-z][a-z0-9_-]*)/check$), measured in-process at tree 1a7db785 (334 paths, 357 operations, API v2.0.0) and pinned for this build in site/content/api-surface.pin.json (path-set sha256 4c03636686101ba2…). Nothing here is typed by hand: the figure moves only when that pinned surface moves. The 14: brand, competitive, compliance, consent, customs, financial, freshness, pii, policy, price, safety, sentiment, spec, subprocessors. Re-derive it against the served document — it counts doors, not paths, so it reads the same whether the document carries a few more or fewer paths than the pin:

bash

curl -s https://api.bluefoxedge.ai/openapi.json \
  | python3 -c 'import json,re,sys; rule=re.compile(r"^/v1/([a-z][a-z0-9_-]*)/check$"); d=sorted({m.group(1) for m in map(rule.match, json.load(sys.stdin)["paths"]) if m}); print(len(d), *d)'

BrandPOST /v1/brand/check · not-yet-validated

Verify brand claims against the reference record

CompetitivePOST /v1/competitive/check · not-yet-validated

Verify competitive claims against reference data

CompliancePOST /v1/compliance/check · not-yet-validated

Verify claims about a brand's regulatory status

ConsentPOST /v1/consent/check · not-yet-validated

Check a consent document against a registered schema

CustomsPOST /v1/customs/check · not-yet-validated

Check N-model agreement on HS/HTS codes for a shipment

FinancialPOST /v1/financial/check · validated: 91.47% match-rate (n=938)

Verify financial claims against EDGAR-grounded facts

FreshnessPOST /v1/freshness/check · not-yet-validated

Check submitted text for stale information

PIIPOST /v1/pii/check · not-yet-validated

Check submitted content for PII with the pinned detector

PolicyPOST /v1/policy/check · not-yet-validated

Verify claims against a brand's stored policies

PricePOST /v1/price/check · not-yet-validated

Verify price claims against recorded product prices

SafetyPOST /v1/safety/check · not-yet-validated

Check content against GARM brand safety categories

SentimentPOST /v1/sentiment/check · not-yet-validated

Analyze sentiment for a batch of texts

SpecPOST /v1/spec/check · not-yet-validated

Verify a product specification claim for a brand

SubprocessorsPOST /v1/subprocessors/check · not-yet-validated

Check observed subprocessors against a disclosed set

PackagePOST /v1/verify/package · a verify door, not one of the 14 check doors above

Verify a software package for registry, vulnerability, and typosquat findings

Keyless reads: the brand catalogue

No key, no account. List the catalogue, then read one brand's definition — the reference every check below is measured against. Executed 2026-09-13; the output is what the door answered, trimmed:

bash

curl -sS "https://api.bluefoxedge.ai/v1/brands?limit=2"
curl -sS https://api.bluefoxedge.ai/v1/brand/nike/definition

json

{
  "data": [
    {
      "slug": "1800-contacts",
      "name": "1-800 Contacts",
      "category": "E-Commerce & Retail",
      "website": "https://www.1800contacts.com"
    }
  ],
  "meta": {
    "request_id": "86a51376-a37a-499c-a43b-a103b307f107",
    "timestamp": "2026-09-13T03:05:10.198224+00:00",
    "pagination": {
      "cursor": "NjAwfDI0IEhvdXIgRml0bmVzcw==",
      "has_more": true,
      "limit": 2,
      "total": 1000
    }
  }
}

json

{
  "data": {
    "slug": "nike",
    "name": "Nike",
    "category": "Unknown",
    "website": "https://www.nike.com",
    "description": "Nike. Just Do It. Nike.com SAQUON APPROVED Change the gam\u2026",
    "positioning": {
      "tagline": null,
      "differentiators": [],
      "market_position": null
    },
    "policies": {
      "shipping": null,
      "warranty": null,
      "return_policy": null
    },
    "pricing": {
      "model": null,
      "range": null
    }
  },
  "meta": {
    "request_id": "64df23b4-ca3e-44cb-9765-e171d0dced7c",
    "timestamp": "2026-09-13T03:05:10.010781+00:00"
  }
}

Example: Brand Verification

bash

curl -sS -X POST https://api.bluefoxedge.ai/v1/brand/check \
  -H "Authorization: Bearer YOUR_KEY" \
  -H "Content-Type: application/json" \
  -d '{"brand": "nike", "claims": ["Nike is a German company"], "depth": "auto"}'

The served envelope (field names from the OpenAPI for POST /v1/brand/check; one result per claim; the verdict is one of accurate · inaccurate · partially_accurate · unverifiable):

json

{
  "data": {
    "brand": "nike",
    "results": [
      {
        "claim": "Nike is a German company",
        "verdict": "inaccurate",
        "confidence": 0.98,
        "confidence_tier": "…",
        "evidence": "…",
        "method": "…",
        "receipt_id": null
      }
    ],
    "brand_definition_version": 1,
    "verified_at": "…",
    "latency_ms": 0,
    "depth_used": "auto"
  },
  "meta": { "request_id": "…", "timestamp": "…" }
}

Example: Policy Verification

bash

curl -sS -X POST https://api.bluefoxedge.ai/v1/policy/check \
  -H "Authorization: Bearer YOUR_KEY" \
  -H "Content-Type: application/json" \
  -d '{"brand": "apple", "claims": ["Apple offers 30-day returns"], "depth": "auto"}'

json

{
  "data": {
    "brand": "apple",
    "depth": "auto",
    "results": [
      {
        "claim": "Apple offers 30-day returns",
        "verdict": "partially_accurate",
        "confidence": 0.72,
        "confidence_tier": "…",
        "evidence": "…",
        "method": "…"
      }
    ],
    "policies_count": 0,
    "verified_at": "…",
    "latency_ms": 0,
    "depth_used": "auto"
  },
  "meta": { "request_id": "…", "timestamp": "…" }
}

Citation Verification

Verify whether a citation (URL + claim) is accurately attributed. Edge checks the source content against the claimed statement and returns a verdict on attribution accuracy.

MCP Scanner

Beyond brand verification, Edge includes a dedicated MCP Scanner that assesses tool safety across 75+ MCP servers — a count recorded April 29, 2026 and not re-measured since (the scanner page dates every figure the same way). See MCP Scanner documentation for details.

Next: Ops Feeds →