The problem this solves
Internal linking is the cheapest SEO lever a site owns and the one most reliably neglected, because maintaining it requires knowing the whole site at once. Every writer links to the handful of pages they remember while the deeper catalog drifts out of the graph entirely. The result on any site more than a year old is predictable: orphan pages that no internal path reaches, money pages that earn a fraction of the links their importance deserves, and a blog whose oldest strong posts hoard authority they never pass along.
The reason it stays broken is that the fix requires a complete picture nobody holds. To place a link well you need to know that the target page exists, that the source page discusses the topic, and that the two are not already connected, multiplied across every page pair on the site. A human editor holds maybe forty pages in mind; the site has four hundred. So internal linking advice stays generic, and generic advice produces the same three navigation links on every page.
This mission holds the whole site in mind at once. It crawls every page, rebuilds the actual link graph from what the pages contain, and emits specific insertions: this sentence on this page should link to that page with this anchor. The deliverable is an editing checklist rather than a strategy document.
How the mission runs
- Crawl the full site. The crawler walks the complete URL inventory from the site mapper and reads every page's content and outbound internal links, so the link graph is rebuilt from what the pages actually contain today rather than from a stale export or a sample.
- Find the graph's failures. From the rebuilt graph the mission identifies orphan pages with no incoming internal links, important pages that are under-linked relative to their role, and pages whose outbound links all point at navigation rather than content. Each failure is quantified, so the fix list has an order.
- Match sources to targets by topic. For every under-linked target, the mission searches the crawled corpus for pages that genuinely discuss the related topic, because a link's value depends on the surrounding relevance. Candidate sources are ranked by topical fit and their own authority within the graph.
- Write the specific insertions. Each recommendation names the source page, the target page, and suggested anchor text drawn from the source's own phrasing, so the editor's job is placement rather than composition. Suggestions avoid over-optimized exact-match anchors as a matter of policy.
- Deliver the ordered CSV. The insertions land as a CSV ordered by expected impact: rescuing orphans first, reinforcing money pages second, general graph strengthening third. Each row is one edit a person can make in under a minute.
The prompt
This is the exact objective the agent receives. Swap the obvious placeholders for your own domain, segment or channel and run it as-is from the console, Slack, or the API.
What comes back
A CSV of specific internal link insertions, each row naming source page, target page, and suggested anchor text, ordered so the highest-impact edits come first, plus a summary of the graph's health: orphan count, the most under-linked important pages, and how the numbers should look after the pass is applied.
Make it yours
- Scope the pass to the blog alone when the goal is routing authority from strong old posts toward the commercial pages they should feed.
- Rerun quarterly and diff the graph summaries: new orphans usually mean the publishing workflow is shipping pages without linking them, which is a process fix worth making once.
- Combine it with the content refresh playbook so pages being rewritten anyway receive their new internal links in the same edit.
Frequently asked questions
Why does internal linking matter this much?
Internal links are how authority and meaning move through a site: they tell search engines which pages you consider important and give crawlers and readers paths to reach them. An orphan page is invisible in practice regardless of quality, and a strong page that links to nothing wastes authority it could be passing to pages that convert.
Will the anchor text suggestions look natural?
The anchors are drawn from language already present on the source page, favoring descriptive phrases over exact-match keyword anchors. The suggestions read as editorial links because they are built from the page's own sentences, and every one passes through your editor's hands before it ships.
Does the mission edit the pages itself?
The default deliverable is the CSV, applied by your editors at their own pace. On sites where the agent has a write path, such as a connected CMS, the insertions can arrive as drafts for approval instead, and nothing publishes without a human accepting it.