Please purchase this course, or sign in if you’re already enrolled, to access the course content.
Lesson overview
This practical turns the method from Lesson 5.5 into a script: from the entity exports, the naming table and the inlinks report, it builds a link plan and a set of link rules in seconds. Claude then finds where each link belongs in the existing text.
Every entity worth linking needs one owner page. The lesson adds an owner column to the naming table and has the script suggest owners where it is blank, marked for checking — and shows why: on the course’s frozen data, an automatic pick sent a general entity to an unrelated glossary page. The script then proposes links from every page that mentions an entity above a salience floor to that entity’s owner, removes self-links, pairs already linked, archives and pagination, caps proposals per page and orders the rest in rounds of twenty with orphan pages first. No API is involved; on the frozen dataset it proposed 438 links across 283 pages.
Claude’s role is narrow and stated precisely: given a page and its rows, quote the sentence where the entity is already discussed and mark the anchor, or report that there is no natural place. It is never asked to write a sentence to hold a link. The lesson compares shipping by editing posts with shipping by link rules applied at render, and closes with a re-crawl that shows the shipped links have dropped out of the plan.
The lesson also covers the practical details that decide whether automatic linking helps or harms: one link per rule per page, no links in headings or existing anchors, a preview before anything ships, and the ability to switch every automatic link off at once if the rules turn out to be wrong.
This practical follows Lesson 5.5 in Module 5 of The AI-powered Entity Audit. It suits learners who have completed the Module 2 and Module 3 practicals and have their entity exports, naming table and inlinks report to hand.
In this lesson you’ll cover: