We're updating our Privacy Policy. The new version takes effect December 14, 2026. Read the updated policy

neural partners

NeuralBot / Site operators

One identity. One reputation.

When NeuralBot fetches your site, it identifies itself with a real user-agent and a link back here. Find out what it wanted, why we sent it, and how to keep it out.

Readiness request identityHTTP header
User-Agent: NeuralBot-Readiness/1.0 (+https://neuralpartners.ai/bot)

One exception: a browser-identified comparison fetch checks for cloaking. Why that request is different →

Two scoped jobsNo open-web crawling

The bot family

Two bots.
Two jobs.

Neither crawls the open web or builds a general index. New variants are documented here as they ship.

NeuralBot-Readiness

On demand
What it does
Scans a submitted URL for AI-agent readiness: structured data, meta and Open Graph tags, robots.txt and llms.txt discovery, cloaking, and more.
Why
Powers the free Agentic Readiness Check on our homepage. Run the check →
Cadence & scope
On-demand only, when someone runs a scan against your domain. A scan makes a bounded set of page, discovery and protocol requests, including a test POST that changes nothing to any agent or commerce endpoint your site declares. Storefronts can also trigger a limited product-page sample and one read-only request for the cart page; it never adds to a cart or submits a checkout. Results are cached for 1 hour per host. It never runs on a schedule.
NeuralBot-Readiness user-agentText
NeuralBot-Readiness/1.0 (+https://neuralpartners.ai/bot)

NeuralBot-Ingest

Scheduled
What it does
Pulls OEM product-catalog data — product details, pricing, images and inventory — from the storefronts, APIs and image CDNs of manufacturers a Neural Showrooms client carries.
Why
Keeps showroom catalogs current with the manufacturers each client stocks.
Cadence & scope
Scheduled pulls of manufacturer inventory and catalog feeds, plus on-demand re-extractions. Operator-configured for onboarded manufacturers only — never open-web crawling or arbitrary domains.
NeuralBot-Ingest user-agentText
NeuralBot-Ingest/1.0 (+https://neuralpartners.ai/bot)

The deliberate exception

The browser baseline.

A Readiness scan also fetches the homepage with a standard Mozilla/5.0 … Chrome browser user-agent. We compare what a normal browser receives with what NeuralBot-Readiness receives to check for cloaking.

That browser-identified comparison fetch comes from the same source and is part of the bounded scan, not a separate crawler. Homepage fetches can be retried after a transient failure; the retry budget and wait are bounded.

We never present ourselves as another company’s crawler.

Your site. Your choice.

How to block us.

Add the token for the bot you want to exclude to your robots.txt. The two variants have different enforcement status.

NeuralBot-Readiness

Honored today

For an uncached scan, we check robots.txt first. If it disallows NeuralBot-Readiness, or all bots with User-agent: *, for the page you submit, we refuse the scan; any other path it disallows is skipped. A cached result may still return until its cache expires after 1 hour.

Block NeuralBot-Readinessrobots.txt
User-agent: NeuralBot-Readiness
Disallow: /

The block applies whether you or someone else requests the scan. A claim to own a site does not override its robots.txt rule: we cannot verify ownership yet. Remove the rule before running a check on your own domain. An ownership-verified override is on our roadmap.

NeuralBot-Ingest

Not yet enforced

NeuralBot-Ingest does not read robots.txt yet. You can declare the token now; until it is enforced, email our bot inbox for an immediate opt-out from catalog ingestion.

Block NeuralBot-Ingestrobots.txt
User-agent: NeuralBot-Ingest
Disallow: /

Request an immediate opt-out →

Verifiability & egress

Recognize us.
Verify with us.

NeuralBot-Readiness signs every request with Web Bot Auth (HTTP Message Signatures, RFC 9421), the method Cloudflare’s Verified Bots program checks. The signature covers @authority and signature-agent, is tagged web-bot-auth, uses Ed25519, and expires 60 seconds after it is made. A redirect is signed again for its new host.

Signed request headersHTTP headers
Signature-Agent: "https://www.neuralpartners.ai"
Signature-Input: sig1=("@authority" "signature-agent");created=…;keyid="…";alg="ed25519";expires=…;nonce="…";tag="web-bot-auth"
Signature: sig1=:…:

The public key is at https://www.neuralpartners.ai/.well-known/http-message-signatures-directory (served by the scanner, not a static file), and that response is itself signed. Its keyid is the key’s JWK thumbprint. One request is unsigned on purpose: the browser-identified cloaking fetch above, because signing it would defeat the comparison. NeuralBot-Ingest does not sign its requests yet.

We don’t publish IP ranges: the scanner runs on changing cloud egress, so the signature, not the address, is the proof. A user-agent string alone proves nothing. Our Cloudflare Verified Bots registration is in progress.

To confirm a request, report traffic that is too aggressive, flag a robots.txt concern, or ask a question, email bots@neuralpartners.ai.

A human reads that inbox and can verify whether the request was ours.