Privacy by design
Privacy notice
FeedGuardrail is a no-charge, browser-local product-feed quality tool. Product-feed data stays on your device.
Effective: September 13, 2026 · Version 2.2
Explicit browser-level choice
Optional, privacy-minimized product analytics
Analytics are off until you enable them on this browser. Turning them off stops future events. Feed scanning, comparison, reports, and evidence verification continue to work locally either way.
FeedGuardrail also honors enabled Global Privacy Control and browser Do Not Track signals. Either signal keeps analytics off even if you select enable. Turning analytics off does not delete earlier records; use the deletion control below for that.
What never leaves your device
Feed files, evidence HTML or JSON files, filenames, file hashes, product IDs, titles, descriptions, prices, URLs, images, row-level validation details, contract and policy contents, and downloaded reports are processed in browser memory. FeedGuardrail does not upload or store them. If you explicitly enable product analytics, only the fixed coarse outcomes described below can be sent.
What optional product analytics records
After you enable analytics, the application creates a random, pseudonymous browser identifier in local storage. It can record page and troubleshooting-guide visits, an allowlisted referral label—including a fixed shared-link label—scan starts and completions, whether a run used a built-in sample or local files, coarse file-size and row-count ranges, aggregate error and warning counts, correction, client-summary, evidence, and change-ledger download actions, Release Gate starts and completions, aggregate resolved/introduced/persistent counts, a fixed comparison verdict, evidence-verification starts, and a fixed verification outcome. It never records the selected release policy or change contract, compared filenames, product IDs, hashes, evidence contents, ledger contents, feed values, file contents, referring URLs, or arbitrary query parameters.
Who is responsible
FeedGuardrail is the service name. Its Alberta-based owner-operator is responsible for the application analytics described here and acts as the privacy lead. Contact hello@feedguardrail.com. The operator’s complete legal identity and service address must be added before paid launch; the current no-charge release does not represent that legal setup as complete.
Reports and controls you download
Correction CSVs, operator change-ledger CSVs, client summaries, change-contract JSON, release-policy JSON, and Release Gate evidence packs are created on your device and are not uploaded to FeedGuardrail. Evidence packs include both local file hashes, the ruleset, exact declared change contract, exact safety policy, and aggregate drift and gate results, but exclude filenames and product IDs by default. Feed rows and raw row-level attribute values are never embedded; fixed normalized availability states can appear only as aggregate transitions. Contract labels, policy labels, allowed field names, and observed field or header names are included because they are part of the reproducible decision, so avoid placing confidential information in them. When explicitly enabled, identifier-bearing evidence includes filenames and up to five product ID samples per aggregate record. The separate operator change ledger intentionally includes product IDs and normalized parsed before/after values for uniquely matched products; it reports missing or duplicate IDs as ambiguous rather than guessing. HTML evidence contains a public FeedGuardrail link so an intended recipient can locally verify the evidence and exact feed files. Review every download and share it only with intended recipients.
Why and how long
Optional analytics are processed with your express choice to understand product usefulness, reliability, and which workflows matter. FeedGuardrail uses a 90-day retention target for application records. Rows older than 90 days are purged during subsequent analytics writes, so they may remain longer while no such activity occurs. The same-browser deletion control below removes records tied to its existing pseudonymous identifier and turns analytics off. When the delayed-event safeguard is available, the server retains a one-way digest of that identifier with a 24-hour expiry; if its capacity is reached, the control keeps the browser identifier for retry and tells you. Expired digests are removed during later analytics or deletion requests and may remain longer during inactivity. Contact messages are kept only as long as reasonably needed to answer the request, meet security or legal duties, and maintain necessary business records.
Service providers and processing locations
The site currently uses OpenAI Sites and Cloudflare infrastructure to deliver pages, protect the service, and store opted-in application analytics. Those providers can process network and operational data—such as IP address, request time, requested path, browser or device metadata, and security signals—under their contracts and policies. Data may be processed outside Alberta or Canada where provider infrastructure operates and may be subject to the laws of those places. FeedGuardrail does not receive these provider logs as part of its application analytics. Read the OpenAI privacy policy and Cloudflare privacy policy.
Sharing and selling
FeedGuardrail does not sell personal information, share it for cross-context behavioral advertising, use it for targeted advertising, or provide it to data brokers. Application analytics are disclosed only to infrastructure providers acting to host, secure, maintain, or improve the service, or when disclosure is legally required. A future asset transfer would require a documented lawful treatment for existing analytics rather than transferring them automatically.
Local storage and cookies
FeedGuardrail uses local storage to remember the analytics choice and, only after opt-in, a random pseudonymous browser identifier and scan count. It does not use application advertising or cross-site tracking cookies. Cloudflare currently sets a short-lived __cf_bm cookie for bot detection and service security; that infrastructure cookie can be set before analytics consent and is not used by FeedGuardrail for product analytics or advertising. Browser controls can remove or block it, although doing so may affect service security or availability. See Cloudflare’s cookie documentation.
Contact and support information
If you email FeedGuardrail, the service receives the sender address, name if supplied, message, and related delivery metadata through its forwarding and mailbox providers. That information is used to answer the request and handle privacy, security, or legal matters; it is not added to product analytics or used for marketing. Do not email confidential feed files, credentials, payment data, health information, or other sensitive personal information.
Security and incidents
Controls include browser-local file processing, HTTPS, restrictive browser security headers, allowlisted analytics fields, request and parser limits, same-origin checks, spreadsheet-formula neutralization, dependency review, and evidence integrity checks. No internet service is risk-free. If a privacy or security incident creates a legally reportable risk, the operator will investigate, contain it, keep required records, and notify regulators or affected people as applicable. Report suspected abuse or vulnerabilities to hello@feedguardrail.com; do not include live customer feed data in the first message.
Your rights and control
Turn off future analytics above. From the same browser, download the application analytics linked to its existing pseudonymous identifier or delete those records below. You may also ask to access or correct personal information, withdraw consent, or challenge the handling of it by emailing the privacy lead. Because FeedGuardrail deliberately does not connect the random identifier to a name or email address, records may be impossible to locate after browser storage is cleared. Depending on where you live, you may complain to the Office of the Information and Privacy Commissioner of Alberta, the Office of the Privacy Commissioner of Canada, or your local data-protection authority.
Children
FeedGuardrail is a professional product-data tool and is not directed to children. Do not enable analytics if you are not old enough to provide valid privacy consent where you live. If the operator learns that a child’s personal information was collected without valid consent, it will be deleted.
Download this browser’s analytics record
Creates a local JSON copy of the application analytics tied to this browser’s existing pseudonymous identifier. It does not create a new identifier.
Delete this browser’s analytics record
This removes the analytics linked to this browser’s current identifier and turns analytics off. It does not affect files or reports because FeedGuardrail never stored them.