Skip to content
nah

Privacy Policy

Last updated: October 31, 2026

Most tools process your files and text in your browser. Short links, dynamic QR codes and link-in-bio pages use hosted storage. Loading the site and downloading its tools also makes network requests, even when the file processing itself is local.

Local tools and browser storage

The browser tools for PDFs, images, media, transcription, developer utilities and text process their inputs on your device. Static QR content is encoded locally. Resume and invoice drafts, data-removal progress and preferences can be saved in browser storage; downloaded AI models can also be cached there. Export important drafts before clearing site data. Data-removal links and request emails take you to third parties and share the information you submit to them.

Hosted links and public pages

Short links and dynamic QR codes store destination URLs, codes and any supplied labels, aliases, expiration dates and UTM fields in Cloudflare D1, with redirect caching in KV. Bio pages store your handle, display name, bio, avatar URL, theme and links. Published profile details and links are public, and external avatars are loaded from their image host.

Management passphrases are sent to the server for creation or verification and stored as bcrypt hashes rather than plain text. A short link can be created without a passphrase; management features require a configured passphrase. Keep your management details somewhere safe.

Visits, clicks and abuse prevention

For passphrase-protected redirects, the redirect worker records the link code, time, country, city, region, device category, referring URL and up to 256 characters of the browser’s user-agent string in Cloudflare Analytics Engine. This click logging is skipped for redirects without a passphrase. Bio pages count views and link clicks; their click logs include country, device category and referring URL in D1.

API requests use IP addresses for rate limiting. Limits on creating and changing links and QR codes count per network and per passphrase using hashes salted with a random value that changes every day; these counters expire after two days and the IP address and passphrase themselves are not stored. The link safety check described below also counts its checks per network, and for IPv6 also per /48 address block, for the current day, in our database, under hashes made with the same daily salt; those counts are deleted by the first check on a later day, and the salt itself expires after two days, after which a stored hash can no longer be linked to an address. Link-abuse reports store the reason and a hash of the reporter’s network address made with the same daily salt, which limits each network to one open report per link per day. When Turnstile is configured, creation APIs verify its token with Cloudflare and include the request IP when available. These records are more than anonymous page counts and can contain information about a visit.

When a short link or dynamic QR code is created or its destination changes, the server checks the destination for phishing and scams. It requests the URL, with any UTM fields, and follows its redirects, identifying itself as the nah.tools link check. The page title, form fields and up to 6,000 characters of visible text are sent to Cloudflare Workers AI for a decision and are not stored. That request can count as a visit to the destination, so links that work only once, such as sign-in or confirmation links, may stop working. If the check finds phishing, a scam, an app installer or too many redirects, the link is not created or changed; otherwise it is.

The service keeps its own log lines in Cloudflare Workers Logs for up to seven days. They are the link check's decisions (event, destination host name, HTTP status and the model's scores, without the rest of the URL) and errors. When a server-side tool fails, such as the PDF and PowerPoint tools offered to AI agents, the log records only the tool's name and the type of error, not the error message or anything from the file. Request URLs, request headers and IP addresses are not logged.

Network services and offline use

Cloudflare provides hosting, storage, redirects and abuse-prevention services. The page template loads Google Fonts. Media processing downloads FFmpeg from unpkg; transcription downloads models through Hugging Face; background removal downloads models from a Cloudflare R2 host. These hosts receive the requests needed to deliver those resources; local processing does not mean there are no third-party connections.

Some local tools may work after their code and required models have loaded, but offline use is not guaranteed. A first download, missing cache or cleared browser storage needs a connection. Hosted links, bio publishing and management require the service to be online.

The application source does not include a Cloudflare Web Analytics beacon or application-set cookies. This does not establish that hosting logs, deployment-added analytics or provider cookies are absent. Their live configuration and retention are not documented in this repository. See Cloudflare’s privacy policy for its handling of infrastructure data.

Deactivation and retention

Deactivating a link, dynamic code or bio page marks its record inactive; it does not erase the database record or associated analytics. Link expiration also does not delete stored data. The source does not define a complete retention or backup-deletion schedule, so we cannot promise that deactivation removes all stored information. Deleting an individual bio link uses a separate delete operation and does not establish deletion of its historical click logs.

Source and contact

You can inspect the implementation and page history in the source repository. Source code describes application behavior, not every live infrastructure setting. For questions, open a GitHub issue; do not post passphrases or private data in a public issue.