How our scanners actually work

Every Digital Empire scanner ships a rules-version string in its response, cites the primary regulatory source (a Federal Register notice, a CFR section, a platform's own developer docs) for every rule it applies, and states in the same UI what the scan does not cover. This page is the umbrella explanation for how PixelProof, EntryProof, and TariffWatch each work today, with links out to each product's own methodology page for the full per-rule detail.

1. Shared standards across all three scanners

The three products solve different problems, but they are held to the same four disciplines, because the customer buying any one of them is the same kind of person (a Shopify merchant, a cross-border seller, a metals importer) who has been burned before by a "compliance tool" that made confident regulatory claims it could not defend. Every scan response, regardless of product, includes:

• A rules-version string. A scan against section232-v1-2026-08-08 today will still be reproducible in 6 months even if we update the ruleset, because the exact string that produced your result is in the response body (result.rulesVersion) and pinned on this page's per-product sections below. If you dispute a scan result later, we can tell you which rules version generated it.
• A primary-source citation for every rule applied. If PixelProof reports that your checkout.liquid tracking will break, the underlying rule references Shopify's own upgrade-to-checkout-extensibility notice — not a third-party blog post interpreting it. If EntryProof reports a CPSC filing gap, the rule references the actual 16 CFR Part 1110 section. If TariffWatch reports a Section 232 exposure, the rule references the actual Federal Register BIS notice cited on that product's methodology page.
• An explicit "what this does not cover" section. Every methodology page lists the classes of decision the scanner deliberately punts on — customs classification advice under 19 CFR 111, GDPR/CCPA legal review under Article 22, individual state consumer-protection statutes, etc. If a scan produces a green result for a case the scanner can't actually rule on, the disclosure is on the same screen as the green.
• No LLM-generated regulatory claims. None of the three scanners uses a large language model to decide whether your store is compliant. Rules are matched by deterministic code against tables encoded from primary sources. LLMs are used for content generation (an explanation string, a suggested next step) but the compliance judgment itself never routes through a model whose reasoning we can't audit deterministically.

2. PixelProof — Meta Pixel & Shopify checkout tracking

PixelProof answers a narrow question: which of the Meta pixels, GA4 tags, and GTM containers on your Shopify store break when Shopify's Aug 26, 2026 checkout-extensibility upgrade removes the Additional Scripts field and checkout.liquid support? The scan does three things: (a) fetches the storefront's public HTML pages via a headless request from our IP (never with a customer's browser session), (b) greps the response for known tracking-loader patterns (fbq(, gtag(, dataLayer.push(, _learnq, etc.), and (c) cross-checks any Meta pixel IDs found against Meta's Graph API public metadata for account activity signals. The scanner does not attempt to place a test order, does not read cookies, does not touch checkout, and does not authenticate against your Shopify admin — every scan is against publicly reachable pages only.

The Aug 26 Shopify-upgrade rule set is encoded from Shopify's Web Pixels API documentation, Checkout Extensibility documentation, and Shopify's own partner-blog deprecation timeline — not from third-party interpretations. Full detail is on /meta-monitor/methodology (rules version, per-check reference, and known false-positive/false-negative modes).

3. EntryProof — CPSC electronic-filing readiness

EntryProof is currently in its v0 "heuristic-only" phase, and we are honest about that on the scanner itself and on /cpsc-efile/methodology. The mandatory CPSC Product Registry electronic-filing rule went live Jul 8, 2026 under 16 CFR Part 1110, requiring specific data fields (importer of record, manufacturer name and address, product identifier, applicable safety rules) to be filed electronically through ACE with the PGA message set. Our v0 scanner does an HTS-chapter to CPSC-applicability mapping (is this HTS classification even subject to CPSC filing?) and a data-completeness check on the fields we can inspect from a store's public product pages.

What EntryProof v0 does not yet do is a full rule-by-rule check against every CPSC safety standard for every product category — that's the M1 rule engine currently in progress and disclosed as such on both the scanner and the roadmap. The v0 heuristic is deliberately conservative: it flags likely gaps, and it does not certify a filing as compliant, because a v0 heuristic scanner is not the right tool to produce that kind of certification. A customs broker or filing agent still owns the final submission.

4. TariffWatch — Section 232 exposure

TariffWatch calculates a Section 232 (steel, aluminum, copper) tariff-exposure estimate for a given list of HTS codes and import volumes. The rules-version string today is section232-v1-2026-08-08 and it references four primary sources: (1) the Aug 6, 2026 Federal Register BIS-14 notice listing the 14 proposed derivative articles, (2) the Apr 6, 2026 Presidential Proclamation establishing the full-customs-value tariff basis with a ≤15% metal-content exemption, (3) the Jun 8, 2026 modification imposing the current 50% rate on aluminum and most steel and copper articles through Dec 31, 2027, and (4) the USITC Harmonized Tariff Schedule for chapter/heading-level rate assignments to chapters 72/73 (steel), 74 (copper), and 76 (aluminum). The full detail — including what the calculator does not cover (exclusions, drawback, first-sale valuation, country-specific rate variations) — is on /tariffwatch/methodology.

5. Data handling & retention

Every free scan across all three products persists the following: the URL scanned, the timestamp, the rules-version string, the pass/fail flags emitted by the scanner, and (if the visitor opted in) an email address so results can be re-sent. Free-scan URL + result data is retained for 30 days, then anonymized to a hashed URL and an aggregated pass/fail counter, per the retention promises on each product's /privacy page. A daily /api/cron/*-data-retention cron enforces this on a schedule visible in vercel.json — the enforcement is not manual, and no team member has admin-panel access to un-anonymize a row past the 30-day boundary.

6. Version control & corrections

Every methodology page in this hub carries a rulesVersion for the underlying scanner and a "last updated" date for the page itself. If a scan today produces a different result than a scan next month, one of two things happened: either the underlying store/import changed, or we shipped a new ruleset (in which case the rules-version string changes and this page is updated to match). If we ever get a rule wrong — a false positive that hurt a customer, a false negative that missed a real gap — the correction lands on /corrections with the date, the affected rule, and the fix. That page is the single audit trail for methodology mistakes; it is not a marketing page.

7. What none of these scanners are

None of the three products is a substitute for a customs broker (19 CFR 111), a trade attorney, a licensed customs house broker, or a CPSC compliance consultant. Every scan result carries a disclosure to that effect in the same UI as the result — not buried in Terms. The value the scanners provide is speed and repeatability on the checks that can be automated deterministically, not a replacement for the professional judgment calls that can't. If a scanner ever presents itself otherwise, that is a bug we want to know about at hello@citationsafe.com.

Related pages

• PixelProof methodology — Meta Pixel and Shopify checkout scanner detail.
• EntryProof methodology — CPSC electronic-filing readiness scanner detail.
• TariffWatch methodology — Section 232 exposure calculator detail.
• Portfolio glossary — cross-product regulatory terminology.
• Portfolio roadmap — what we've shipped, what we're shipping, what we won't.
• Portfolio changelog — weekly release notes from git history.
• Corrections — every methodology mistake we've made and fixed.
• Security & disclosure — how to report a scanner bug.