Related Posts — AddThisFeature

Show a few genuinely relevant articles at the end of a post so readers keep going.

moderate Content & Publishing

What it adds

A block of suggested articles below each post, chosen from tags, topic, and recency.

What your agent is told to do

1
Score candidates from signals the app already has: shared tags from Labels and Tags or AI Topic Tagging, the same category, and closeness in publish date. Combine them into one ranking rather than picking the first tag match.
2
If the app already has Semantic Search or an AI Recommendation Engine, use it as the similarity source and treat tags as a tiebreaker. Do not build a second relevance system beside the one that already exists.
3
Compute relations in the background and store the result. Ranking the whole archive on every page view will be the slowest part of the article page.
4
Give the block a graceful fallback: when nothing scores well, show recent posts from the same category, and when there is still nothing, render nothing rather than an empty heading.
5
Do not suggest posts purely by popularity. A view-count ranking sends every reader to the same three evergreen articles and the rest of the archive never surfaces.

Edge cases it handles

A newly published post has no tags and no signal to match on. Fall back to the same category and to recency rather than showing an empty block on every new article.
The current post must be excluded from its own suggestions. It scores highest against itself on every similarity measure.
Drafts, scheduled posts, and posts behind a paywall or permission check must never appear as suggestions. Filter by the same visibility rules the index page uses, not just by a published flag.
Without a diversity rule, a handful of heavily tagged evergreen posts will appear under every article. Cap how often any single post can be suggested, or mix in a slot chosen from less-surfaced content.
Retagging a large archive triggers recomputation for every affected post. Queue that work in batches so a bulk tag edit does not stall the site or the job queue.
A deleted or unpublished post must drop out of stored relations immediately, not on the next scheduled recompute.
Suggestion links need their own tracking or link attribution if the app measures navigation, otherwise there is no way to tell whether the block works.
Reserve the block's height before the suggestions render so the end of the article does not jump under a reader who has just scrolled there.

Definition of done

Every published post shows relevant suggestions or nothing at all, never an empty block.
A post never suggests itself.
Drafts, scheduled, and access-restricted posts are absent from all suggestions.
No single post appears in more than a capped share of suggestion blocks across the archive.
Relations are precomputed and the article page does not run a ranking query on request.
A bulk retag recomputes affected relations in batches without blocking publishing.
The block occupies reserved space and does not shift the page when it loads.
The feature matches the existing design system.
No existing functionality is broken.

Related features

Reading Progress Bar

Press Kit

FAQ Section

How it works

1

Copy the link

Grab the Markdown instruction URL for this feature.
2

Give it to your AI

Paste it into Claude Code, Cursor, v0, Lovable — whatever you build with.
3

It inspects, then implements

Your agent reads your existing app first, then adds the feature to fit it.

Works with your stack

These instructions are written to adapt. They tell the agent to detect your framework, match your existing design system, and reuse what you already have — rather than assuming a particular stack.

Need it tighter than that? Customize the feature and tell it exactly what you're running.