# Seed URLs

Canonical HTML: /atlas/seed-urls
Atlas version: atlas-2026-07-28-v1
Category: Discovery and crawl frontiers

## Definition

Seed URLs are the starting addresses placed into a crawler before link extraction begins.

## How it works

A seed can come from a prior crawl, a sitemap, a manual submission, a feed, or a link found on another host. Its quality determines which connected region of the web becomes reachable.

## How to validate it

Record the entrypoint separately from later discoveries, then compare which pages become reachable from each seed set under the same page and depth budget.

## Common failure

Treating every visited URL as a seed hides the actual discovery path and makes coverage comparisons impossible.

## Related topics

- [The crawl frontier](/atlas/crawl-frontier)
- [Internal link discovery](/atlas/internal-links)