Ask an AI assistant which WordPress plugin to install and you may receive a confident answer, complete with reasons and familiar names. What the answer usually cannot establish is just as important:
which potentially relevant plugins were never considered?WordPress has a large plugin ecosystem. Finding a plugin by name is relatively easy. Discovering the best functional match when you only know your problem is much harder.
WP-Repo is an independent project being developed around that distinction.Its starting point is a defined set of WordPress.org plugins, with the goal of making their documented capabilities easier to discover by requirement rather than by popularity alone. The interactive discovery service is not yet available.
A good answer can still miss the right candidate
Consider a WooCommerce shop owner who asks:
I need distance-based delivery fees, but I do not want to depend on another paid external service.
A plugin may be popular, well reviewed and easy to recommend, yet fail the constraint about an external service. Another may describe the required capability in unfamiliar terms and barely appear in public discussions.
The difficult question is not just whether an assistant can explain its recommendation. It is whether it
looked across an appropriate candidate set before making one.
The size of the low-adoption group matters
In a November 2026 analysis of locally stored WordPress.org-derived data, WP-Repo selected
26,851 plugin records whose declared
Tested up to value began with
7..Among those records,
13,251 (49.35%) had an
active_installs value of
0, which WordPress presents as
fewer than 10 active installations.
That is not a count of plugins ignored by AI. It is a measurable reminder of how large a group can have almost no positive
adoption signal to support a conventional recommendation. Whether some of those plugins meet a user's needs is a separate, functional question.
Where AI recommendations can reinforce the gap
Reviews, install counts and broad public coverage can make a product easier to discuss and its adoption easier to assess. They cannot establish whether its particular features satisfy your requirement.
If an AI assistant draws candidates primarily from familiar or readily retrieved products, then uses their stronger reputation signals to justify its choice, an already visible plugin can gain another advantage. A specialist plugin may never reach the comparison stage.
This is a
plausible reinforcing mechanism, not a measured claim that all chatbots exclude low-installation plugins. More precise prompts and live web search can help, but neither guarantees a complete, current plugin inventory.
What a more useful discovery process should do
First establish what the user needs. Then find candidate plugins in a
defined, current-enough inventory and examine their documented functions and constraints. Only then assess adoption, maintenance, reviews and other signals that affect confidence in installing one.
A README is a source of functionality claims, not proof that code performs flawlessly. Likewise, an installation count is an adoption indicator, not a functional test. A useful result must distinguish these kinds of evidence and say what remains unknown.
WP-Repo is working toward that approach. It will not be a security certification, a guarantee of compatibility or an endorsement of every relevant-looking plugin.
Relevance should determine who gets examined; the available evidence should determine how confidently each candidate can be recommended.WP-Repo is an independent project and is not affiliated with or endorsed by WordPress.org, the WordPress Foundation or Automattic.