About SojournerBot

If you found this page from your server logs: SojournerBot is the crawler behind Parallax, an independent search engine made by Interdiction Studios. It visits briefly, reads a page the way a browser would, follows the links it finds, and moves on. What it reads goes into one search index that shows results as they are, with no ads and no paid placement.

How it identifies itself

Every request carries this user agent, and the crawler answers to the robots.txt name SojournerBot:

SojournerBot/1.0 (+https://parallaxsearch.com/bot; Interdiction Studios)

Checking that it is really SojournerBot

Anyone can copy a user agent. What they can’t copy is the signature: every request SojournerBot sends, robots.txt included, is signed with Web Bot Auth (HTTP Message Signatures, RFC 9421, using Ed25519), and only we hold the key that makes those signatures. Each request carries three headers:

• Signature-Agent: "https://parallaxsearch.com"
• Signature-Input: what is signed (the host the request was sent to, and the Signature-Agent header), when, and with which key: keyid="x8NQ3Ym_T0epFBgiTOLLU36e9kgvFOy-ONXEYYUF7gc", alg="ed25519", tag="web-bot-auth"
• Signature: the signature itself

To check a request:

1. Take our public key from https://parallaxsearch.com/.well-known/http-message-signatures-directory: the one whose key ID matches keyid. The directory is signed with the same key.
2. Check the signature over what Signature-Input lists.
3. Check the time: a signature is good for five minutes after it was made.

Cloudflare and other services that support Web Bot Auth can check this for you. Most access logs don’t record these headers by default, so checking them yourself may mean logging them first.

We don’t publish a list of IP addresses. The signature doesn’t depend on where the crawler runs, so it is the more reliable check.

How it behaves

• It reads robots.txt before anything else and obeys it, including crawl delays.
• One request at a time per site, with a pause between requests; it never runs many fetches against one site at once.
• It fetches HTML pages only, up to a modest size, and never submits forms, logs in, or runs scripts.
• A page it has read is revisited rarely: after a week at first, and less often the less it changes.
• It stores the page’s text and its links, and nothing about anyone who visits your site.

Keeping it out

You don’t need to ask anyone. Any of these works, and takes effect the next time the crawler looks:

# robots.txt: keep it off the whole site
User-agent: SojournerBot
Disallow: /

# ...or off one part of it
User-agent: SojournerBot
Disallow: /private/

A page with <meta name="robots" content="noindex"> is dropped from the index the next time it is read, and so is a page that starts answering 404 or 410.

Getting in touch

To have something removed sooner, or to tell us the crawler is misbehaving, write to bot@parallaxsearch.com. A person reads every message.