# Origin access and crawler allowlisting - Agent Delivery Network | Agent Delivery Network

<!-- adn:geo-lead -->
> Let ADN ingest reach your HTML: Web Bot Auth, x-adn-crawler-token, ADN-Corpus/1.1, Test fetch, token rotation, and Cloudflare IP ranges as a last resort.

<!-- /adn:geo-lead -->

Back to Help ADN ingest fetches your origin HTML to build Markdown, discovery files, and catalog indexes. Those corpus requests must not be blocked by your firewall, bot management, or host rules.
## What ADN sends on corpus fetches
 1. Web Bot Auth - HTTP Message Signatures ( Signature , Signature-Input , Signature-Agent ) using Ed25519 keys. Public keys are published at /.well-known/http-message-signatures-directory on the marketing site and on your gateway host when configured.
 2. Per-domain header - x-adn-crawler-token (secret you rotate in the dashboard).
 3. User-Agent - ADN-Corpus/1.1 (+https://agent-delivery.network/bot) .
 Canonical crawler page: bot info page .
## Dashboard workflow
 On **Domains → Connect → Origin access** (pick your domain first):
 1. ADN detects your origin stack (Cloudflare, Vercel, Netlify, WordPress, and others) and shows tailored steps.
 2. Create or Rotate crawler token if you need a fresh x-adn-crawler-token . Copy the token when shown; it is not shown again.
 3. Click Test fetch . A success result means ingest can read a sample page. Failures include a title, detail, and fix hint.
 4. If ingest was blocked, a banner may show the last failure reason until you fix allow rules.
## Allowlisting by stack
 Follow the copy blocks in the dashboard panel. In general:
 • Cloudflare - Prefer WAF Skip or allow rules on verified bots, Web Bot Auth ( cf.bot_management.signed_agent when approved), and x-adn-crawler-token . Do not rely on User-Agent alone.
 • Vercel / Netlify - Exempt corpus paths or allow the headers above in Firewall settings.
 • WordPress / generic - Allow the User-Agent and token header on routes ingest needs.
## Token rotation
 Rotate when the token may have leaked, a contractor leaves, or you change firewall vendors. After rotation, update every allow rule that checks x-adn-crawler-token . Old tokens stop working immediately.
## Cloudflare IP ranges (last resort only)
 Ingest egress uses Cloudflare Workers addresses shared with other customers. IP allowlists are brittle. Prefer Web Bot Auth and x-adn-crawler-token . If you must use IPs, see the link on the bot info page to Cloudflare's published ranges.
## Related
 • Ingest and crawl status
 • Getting started
 • Connection methods compared
 Ask ADN