GetMarkdown
← GetMarkdown converter

Privacy

What stays in your browser and what does not

Files converted in your browser stay local. Web page conversion and Pro PDF conversion run on our servers. This page explains the data used by each path, accounts, API and agent keys, payments, and analytics.

Effective November 5, 2026

Free browser conversion

Supported free-tier files are processed locally in the browser. GetMarkdown does not intentionally transmit the selected file contents to its servers for that conversion path.

The browser downloads runtime and package resources needed for conversion. Ordinary network requests for the website and those resources can include standard technical information such as IP address, user agent, requested URL, and request time; they do not need the contents of the selected document.

Accounts

The Pro website requires a free account. Sign-in accounts are provided by Clerk, which stores your email address and sends the one-time sign-in codes. GetMarkdown keeps your email address, credit balance, purchase and plan records, and your PDF job records (file name, size, page counts, status, and credits charged). Paid web page conversions also create credit-spend records, including the source hostname. Free browser file conversion and the free web page allowance do not require an account. Agent keys use a separate account without Clerk sign-in, described below.

Web page conversion

When you submit a web page address through the website or API, our server fetches it with a User-Agent identifying GetMarkdown/1.0. The destination site receives that server request. We do not send your browser cookies, API key, or visitor IP to the destination. Cloudflare caches the converted Markdown, page metadata, and final URL for up to one hour. The cache is shared across callers requesting the same URL.

When browser rendering is used, we send the public page URL to Cloudflare Browser Run to load page content and resources. Those requests may use Cloudflare's browser identity. We do not forward your browser cookies, sign-in session, or API key to the destination. Browser-rendered and ordinary extraction results use separate caches.

To enforce the free daily allowance, the website forwards your visitor IP to our API. For signed-out requests, we store a SHA-256 hash of that IP in Cloudflare D1, together with the UTC day and request count. IPv6 addresses are grouped by their /64 network before hashing. For signed-in requests, the allowance record uses your account identifier instead.

The allowance resets at midnight UTC. The hourly cleanup deletes records for earlier UTC days, normally on the first scheduled run after midnight. These counters store the hash or account identifier, day, and count, rather than the submitted URL or raw IP. Ordinary hosting and network requests can still include IP addresses as described above.

Website conversion

Website conversion requires an account session or API key. We send the starting public URL and your crawl options to Cloudflare, which discovers and converts linked pages. If you enable JavaScript, its browser loads page resources. Visitor credentials are not forwarded to those websites.

Cloudflare stores crawl results for up to 14 days. GetMarkdown makes its exports available for 48 hours after completion, with hourly cleanup removing expired exports. We keep the starting URL, options, page outcomes, conversion status, timestamps, and credit reservation, charge, and refund records separately for account history and billing. These metadata and billing records do not have a fixed automatic deletion period.

API keys

The full API key is returned only when it is created or first claimed. We store a SHA-256 hash of the key, its name and prefix, its account identifier, creation and last-used times, and any revocation time. Credit-spend records identify the key used for a charge. We also keep an account request counter to enforce API rate limits. Key and spending records do not have a fixed automatic deletion period.

Agent keys and purchases without sign-in

An agent key purchase creates a separate account whose credits are accessed through the key. GetMarkdown does not collect an email address for this account or send a sign-in email. Stripe collects payment details and an email address during checkout. We keep the account, credit and purchase records needed to apply payments and refunds.

We store purchase session records in D1, including a SHA-256 hash of the private polling token, account and key identifiers, selected pack, credit amount, Stripe checkout request and session identifier, status, and timestamps. Checkout sessions are created with a 60-minute payment window. That window is not a deletion deadline or a deadline for claiming a paid key.

New purchases and top-ups share a limit of 5 checkout attempts per IP hash per hour. The website forwards the visitor IP to our API; we hash it after the same IPv6 grouping used for web page allowances. These counters store the hash, hour, and count. Hourly cleanup deletes counters from earlier hours and checks unpaid sessions with Stripe before expiring them. If payment status cannot be confirmed, reconciliation is retried later. Unpaid agent accounts are removed only when they have no remaining active session, key, credit history, jobs, or ledger or subscription records. Purchase session records themselves have no fixed automatic deletion period.

MCP

The MCP convert_url tool sends the requested URL and caller IP to our conversion API, along with an API key if supplied. It uses the same web page processing, cache, allowance, and credit-spend records described above. Website conversion tools use the website processing and retention rules above. PDF conversion tools upload the PDF to our service and use the same processing and retention rules as Pro PDF conversion.

Pro PDF conversion

Pro is server-side. The selected PDF is uploaded to GetMarkdown's conversion service on Cloudflare (Workers and R2 storage), where it is inspected for a quote, converted, and packaged as a download.

When a PDF has scanned pages that need OCR, when the layout engine cannot extract usable text from it, when it has more than 300 pages, or when you request images, the PDF file is sent to Datalab, which converts it (including text recognition) and returns the result.

  • Jobs that have not started become eligible for upload deletion 2 hours after job creation. Hourly cleanup removes them on its next run.
  • Finished downloads expire 48 hours after the job finishes. Hourly cleanup then deletes the upload and output files.
  • Account, credit, and job records needed to show your history, reconcile a payment, or diagnose a service failure may be retained separately from document contents.

Payments

Stripe handles checkout, subscriptions, the billing portal, and payment-card processing for credit packs and plans, including agent key purchases and top-ups. Credits can pay for PDF and web page conversion through the website, API, or MCP. GetMarkdown does not receive or store full payment-card details. Stripe processes payment information under its own privacy terms.

Service providers

  • Vercel hosts the website.
  • Clerk provides accounts and email sign-in.
  • Cloudflare (Workers, Browser Run, R2, D1, and the Cache API) processes and stores PDF uploads and outputs, rendered web pages, website crawl results, account and credit records, API keys, purchase sessions, and usage counters.
  • Datalab performs OCR on scanned pages and converts PDFs the in-house engine cannot handle.
  • Stripe processes payments and billing.

Site analytics

GetMarkdown uses Vercel Analytics and Simple Analytics to understand aggregate site usage. Analytics may receive ordinary page-request and device information. File contents are not intentionally included in analytics events.

Your choices

Use browser file conversion when you need local processing. Submit only public web page addresses you are authorized to access, without private information in the URL. Do not submit a PDF through Pro, the API, or MCP if you are not authorized to send it to our servers and the providers listed above, or if a contractual, legal, or organizational policy prohibits that processing.

Contact

Questions about privacy, your account, web page conversion, API or agent keys, MCP, or a PDF job can be sent to aark@getmarkdown.com. Never send a full API key or private polling token by email.