Privacy

Last updated 2026-11-14. Company: SwarmInc. Contact: SwarmInc@proton.me. Jurisdiction: not stated.

This page says what the company's own software records about a visitor or a caller, read from that software. The platforms a product is bought on (RapidAPI, Apify) keep their own records of your account and payments under their own privacy policies; SwarmInc never sees a card number or a payment.

What the site counts

Every page carries a 1×1 image that the company's own server answers (/p?u=<page>). It counts one view of that page path in a day's bucket, split into people and robots by the words in the browser's user-agent string. A link from a page to a RapidAPI listing goes through the same server (/go/<target>), which counts the click under the page it came from and redirects to the listing named in the site's own configuration. No cookie is set, no IP address and no user-agent string is kept: a view is a number in a bucket. A client's address is held in the server's memory, salted and hashed, for one hour only, to count at most 30 views from it an hour; it is never written. The counter keeps its counts 90 days (swarmctl/sitecount.py). The counts of arrivals by the referring site's category, page and day (below) are also copied into the company's own distribution record and kept 120 days there, to compare a page's title and description with the arrivals it gets; they are numbers per page and day, never about a visitor (swarmctl/distribution.py).

The referring site's category, and nothing else about where you came from: with JavaScript on, a script in the page reads the referring address in your browser, or a link's utm_source when there is none, and sends only its category, one of baidu, bing, brave, chatgpt, claude, copilot, direct, duckduckgo, ecosia, gemini, google, other-site, perplexity, yahoo, yandex (&s=<category>); a visit from another page of this site sends none. The referring address itself never leaves your browser and is never kept. That category is counted with the page and the day. A link the company placed elsewhere may carry a tag naming that channel (?src=<tag>); it is counted with the page only when it is one of the company's own tags. Without JavaScript, a view is counted with no category.

The support desk

The support desk (/support) takes a question without an account or an e-mail address. A ticket holds the product you chose, the question you typed (at most 3,000 characters), the request id an error answer gave you if you entered one, and the time. It holds no address, cookie or user-agent string; to limit how many tickets one visitor opens (an hourly and a daily limit), the server keeps a salted hash of the address and the times it asked in memory for a day, never on disk. The ticket's page is the only channel back: anyone with its address can read it, and nobody without it can. The answer is written by an automated assistant from the product's documentation and the server's own record of your request, and every answer says so; a question about billing, a refund, an account, privacy or a legal matter is not answered by the assistant but marked for a person, and the first 400 characters of such a question are copied to the owner's queue. The desk's retention setting is 90 days; no ticket is deleted sooner (swarmctl/supportdesk.py).

API calls

A call to an API reaches the company's server through RapidAPI's gateway. The server counts calls per product and per plan, and keeps a per-subscriber rate table in memory under RapidAPI's subscriber id (or the caller's network when the call did not come through RapidAPI). It keeps each subscriber's calls by product, plan and day under a salted hash of that id, never the id or a key, for this calendar month and the 1 before it (2 months in all). A call that carries one of the company's own channel tags (src=<tag>, set by its workflow templates) is also counted under that tag; the same hashes are kept there for as long, by month, and after that only each month's number of distinct callers stays. The tag is removed before the product reads the query. When a call fails, the server records an incident: an id it gives you, the product, the status code and its own reason — never the caller, a parameter or a header (swarmctl/apiserve.py). The rows a product returns come from the public source its page names and are not records about you.

No sale of data

SwarmInc sells data products that read public sources. It does not sell, rent or share records about visitors, callers or ticket writers; the records above are kept for the service's own operation and shown on the company's internal status report.

Your questions

A question about these records, or a request to remove a ticket, goes to SwarmInc@proton.me or to the support desk, where it is marked for a person.