Screen by Screen: the IndexMesh Setup Guide WordPress Needs for AI Discovery, Robots, and Schema & Entities

The IndexMesh setup guide WordPress install and activation steps

Install IndexMesh for WordPress from the WordPress plugin directory and activate it. No wizard opens on activation. Setup is offered instead, from three dismissible entry points: the Get started link on the Plugins screen, a notice after activation, and the welcome panel on the IndexMesh Home screen. You can dismiss all three and configure each screen by hand, which is what the rest of this guide walks through.

The Home screen also carries a Compatibility tab that shows which SEO provider it detected on your site. Check that tab first if you already run Yoast, Rank Math, AIOSEO, or SEOPress, it tells you which provider IndexMesh for WordPress is enriching rather than duplicating.

Menu labels match the plugin's own screens. The llms.txt builder sits under AI Discovery, which is IndexMesh's name for the public file and the AI-facing site summary around it. The file is llms.txt. The screen is AI Discovery. A reader will meet both words and they are not interchangeable.

The four steps, done by hand

If you skip the guided setup, the same ground is four steps:

1. Robots. Open Robots > Crawler controls and turn on the preset groups that fit your site. After saving, use Preview to see the exact robots.txt output.
2. llms.txt. Open IndexMesh > AI Discovery and work through the guided sections. Fill from site info prefills detected values from your WordPress settings and Organisation schema.
3. Schema. Open Schema & Entities > Your business and add your details. Author details live on each user’s WordPress profile.
4. Verify. Open Status and confirm your public routes respond, then Home > Compatibility to confirm which SEO provider is active.

Set up the AI Discovery screen

The builder lives at IndexMesh > AI Discovery and has three tabs.

Fill from site info prefills detected values from your WordPress settings and Organisation schema, so you review rather than retype.

Per item control

Posts and pages carry an IndexMesh llms.txt control in the editor. You can exclude an item, or pin it above automatically selected content. Pinned items render first.

Advanced source rules cap the maximum number of automatic items and toggle automatic pages and recent posts independently.

What is never included

Output omits private items, drafts, password-protected content, anything marked noindex, anything explicitly excluded, and any item without a safe HTTP, HTTPS or relative URL.

Generated excerpts are cleaned before output, so theme tails such as "Continue Reading" do not end up in the public file.

Markdown is sanitised. HTML is stripped, link labels are escaped and script-like URLs are rejected, so admin markup cannot reach the public file.

The routes

The manifest exposes no write tools and is not a server of any kind. It is a static description of where the other routes are.

If another plugin already produces llms.txt

The plugin detects the conflict and defers by default. It reports which source appears to own the surface and never overwrites a file it did not create.

To take over, use the Settings tab. The plugin backs up the existing file, imports the content it can map into the builder, replaces the file with its own version and marks it as IndexMesh-owned. If that marker is later removed, it treats the file as foreign again rather than silently reclaiming it.

A root conflict applies to /llms.txt only. The .well-known alias and the REST representation keep serving.

Not available in this release

llms-full.txt. agents.json. Per-language output: WPML and Polylang are detected, but the plugin publishes one file, in the site's default language. A migration importer for another plugin's file.

Configure robots governance presets

Crawler controls live at IndexMesh > Robots and compose a managed block that is appended after your existing robots.txt output.

The preset groups

WordPress. Shop pages. Sitemaps. Search engines. Social. AI. Files. Scrapers.

Preset groups render before append-only custom directives, and Content Signal directives render deterministically after each user-agent group inside the IndexMesh block.

Staging, local and headless sites

The plugin does not write Disallow: / for you. If it detects a staging, local or headless environment, the top of the screen prompts you to turn on Site Access Protection.

That feature is off by default and blocks crawlers only after you turn it on.

Content Signals

Three directives: search, ai-input and ai-train.

They are advisory crawler hints. Setting one to no states a preference and does not stop a request arriving. No legal enforceability is claimed. RSL licensing emission is not in this release.

Physical file or virtual route

Virtual is the default. The plugin composes its block through the WordPress robots_txt filter, after whatever your SEO provider already contributes.

If a physical robots.txt exists at the web root, your server may serve it before any WordPress filter runs. The plugin reports that conflict rather than claiming control it does not have.

Static export reports one of five states.

Export refuses to write unless the mode is managed-file.

The plugin never deletes a physical robots.txt, including on mode change, module disable and plugin deactivation. Turning managed-file mode off leaves the file in place for you to remove.

Importing existing rules

Import reads the physical file and the served output. Rules from other SEO plugins are read from the served output only. The plugin never reads another plugin's private settings tables.

Recognised lines map into preset groups. Unmatched lines are preserved as additive advanced directives. The decision is recorded as pending, imported or skipped.

No daily re-export

There is deliberately no re-export cron. Saves and settings changes sync the file at write time. A daily job would mask an external edit you should see.

Add your Schema and Entities details

Organisation schema is managed at IndexMesh > Schema & Entities, and author schema on each user's native WordPress profile.

Organisation

The Organisation tab covers identity (name, type, brand, legal and alternate name, slogan, URL), a media-library logo picker, description, founding date, employees, knowsAbout, sameAs, contact point, postal address, founder Person, sub-organisations, physical offices with full addresses, the areas you serve, organisation identifiers and publishing principles.

Identifiers: vatID, taxID, iso6523Code, duns, leiCode, naics.

Publishing principles: publishingPrinciples, ownershipFundingInfo, actionableFeedbackPolicy, correctionsPolicy, ethicsPolicy, diversityPolicy, diversityStaffingReport.

A completion notice reports which recommended fields are still empty. Detected Yoast, WordPress and ACF values preload into empty fields for review before saving.

Authors

The Authors tab is a per-user table showing enabled state, job title, and knowsAbout and sameAs counts.

Author fields live on the WordPress user profile, in the IndexMesh Author Schema section, collapsed until enabled: headline, honorific prefix and suffix, job title, employer, headshot, short bio, rich-text full bio, languages, awards, knowsAbout, additional profile URLs, identifiers including Wikidata, ORCID and alumni, and optional author-archive SEO title and description.

Author sameAs priority order: website, LinkedIn, Wikidata, Wikipedia, Google Business Profile, social profiles, then ORCID.

Nothing is emitted until you enable it. Organisation output is off until an operator enables it and saves a name. Author schema is off per user until that author is enabled on their profile.

What gets emitted, and when

The decision depends on what else is running.

Noindex, private, draft, pending, trashed, password-protected, excluded post types and per-post exclusions suppress both output and enrichment.

Open Schema & Entities > Schema Health to see the current decision.

FAQ schema

Per post, opt-in. It is eligible when a local FAQ source is detected and suppressed when another plugin's FAQ block is already on the page.

Google retired the FAQ rich result, which stopped showing on 7 May 2026 with the documentation removed on 15 June 2026. FAQPage remains valid Schema.org markup and the plugin still ships it for answer engines, not for a rich result.

The author shortcode

[indexmesh_author_info] renders author display data from IndexMesh author fields, falling back to Yoast user schema and then WordPress core.

Additional shortcode tags can be registered through the indexmesh_author_info_shortcode_tags filter.

Not available in this release

QAPage, HowTo, ClaimReview and Speakable are not public features. There is no durable findings history: the findings table is deliberately not installed, so lint results do not persist across sessions.

Verify the setup with Status and WP-Cron

IndexMesh > Status is the health report, and the first place to look when something appears wrong.

What it checks

Public files and routes. Live checks against robots.txt, llms.txt, the .well-known mirror, your sitemap and your homepage schema. Each row shows OK, Attention or Problem with a plain-language message.

The sitemap check follows redirects with a bounded hop limit, so a redirect loop fails clearly instead of hanging.

SEO plugin compatibility. Which provider is active, and whether more than one is active at once. The plugin surfaces that state and does not resolve it for you.

Results cache for fifteen minutes. Use the recheck action to force a fresh run.

The Support report

A plain-text snapshot of the plugin version, the check timestamp, the active SEO provider, the sitemap path and the route results.

It is meant to be copied into a support request or bug report. Nothing is sent anywhere automatically.

If a row shows Attention or Problem

Public files and routes. Confirm the route resolves in a normal browser tab first. A check can fail if your host blocks the site from requesting itself, which some firewalls and security plugins do.

SEO plugin compatibility. If more than one SEO plugin is active, deactivate the one you are not using. Running two providers at once causes duplicate or conflicting schema independently of IndexMesh.

Schema. Open Schema & Entities > Schema Health for the current emission decision.

Roles and permissions

The plugin uses its own WordPress capabilities, so you can delegate content work without granting full administrative access.

Administrators always retain full access.

Enabling a role preset

1. Sign in as an administrator.
2. Open IndexMesh > Settings.
3. Find Access control.
4. Read the description for each role.
5. Turn on only the roles that need access.

Presets are off until you enable them. Turning one on grants that role the capabilities described and nothing else.

What IndexMesh for WordPress stores and what leaves your site

By design, the IndexMesh WordPress plugin does not send data to IndexMesh, or to any third party.

External requests

Two call sites make outbound HTTP requests, and both target your own hostname only.

Everything else is local. Bot matching, provider detection, robots governance, llms.txt, discovery routing, conflict detection, schema generation and citation visibility make no network requests at all.

What is stored

Options prefixed indexmesh_, covering module enablement, robots policy, llms.txt configuration, exclusion and pin lists, conflict policy, notice dismissal state, and the Organisation and author schema you enter.

Author fields live in native WordPress user meta.

No audit log and no findings table are created in this release. The indexmesh_audit_log and indexmesh_findings tables are dormant. No request metadata, user-agent text or IP data is written anywhere.

Citation visibility stores nothing. No citation events, no IP addresses, no request logs, no provider keys, no scoring data. It exposes read-only signals derived from the bundled bot taxonomy.

Background activity

Two daily WP-Cron events, both on by default.

Both are cleared on deactivation and do not run again until the plugin is reactivated.

Integrations that are wired and not switched on

Google Search Console and the Redirection plugin are status seams only. No OAuth token is stored and Google is never called. They exist so surrounding work has somewhere to report state.

Troubleshooting

Most reported problems are one of four things.

Developer reference

The Core layer is read-only and performs no provider writes.

Names

Read-only REST routes

GET /wp-json/indexmesh/v1/status
GET /wp-json/indexmesh/v1/citations/signals
GET /wp-json/indexmesh/v1/schema/diagnostics
GET /wp-json/indexmesh/v1/llms

Registries

Module Registry, IndexMesh\Core\ModuleRegistry. Modules implement ModuleInterface and register by lazy factory callback, so a disabled module is never constructed. Register through the indexmesh_register_modules action rather than modifying the bootstrap.

SEO Provider Registry, IndexMesh\Provider\SeoProviderRegistry. Adapters expose read-only status and conservative capability flags for titles, meta descriptions, canonicals, noindex, schema graph support and redirects. Core reads no private provider tables and requires no SEO plugin.

Schema enrichment filters used

Providers receiving the supplemental block are filterable via indexmesh_supplemental_node_providers.

Capabilities

indexmesh_view_schema, indexmesh_edit_schema, indexmesh_manage_entity_store, indexmesh_eeat_can_review, indexmesh_ymyl_domain.

Common tasks

Five things people ask for by name, each answered in one place.