strapi-global-search

Spotlight-style global search for the Strapi v5 admin: every collection type and single type, from a dedicated page or a Ctrl/Cmd + K palette.

Install it, enable it, and start Strapi. It introspects your content types at boot and never needs to know your schema ahead of time.

Strapi 4: use strapi-global-search@1.x. This major (2.x) is Strapi 5 only.

Features

• Searches everything — every api:: collection type and single type, including text fields nested inside components.
• Two surfaces — a full search page under the plugins menu, and a command palette available from anywhere in the admin.
• Predictable ordering — results are ranked id → name → other fields → nested fields (see below), never by an opaque relevance score.
• Respects your permissions — an editor only ever sees content types they can read in the Content Manager.
• i18n and draft/publish aware — searches every locale by default and finds drafts.
• No index, no extra services — plain database queries, nothing to keep in sync.

Installation

npm install strapi-global-search
# or
yarn add strapi-global-search

Enable it in config/plugins.js or config/plugins.ts:

module.exports = {
	"global-search": {
		enabled: true,
	},
};

Then start your Strapi app (npm run develop). There is no extra build step.

This package has no preinstall / install / postinstall scripts and does not use git or remote URL dependencies. On npm v12 you do not need an allowScripts entry for strapi-global-search.

Result ordering

Ties break on an exact-case match, then the shortest matching value, then the most recently updated entry.

Configuration

Everything is optional and can be changed at Settings → Global Search → Configuration, or pinned in config/plugins.js / config/plugins.ts:

module.exports = {
	"global-search": {
		enabled: true,
		config: {
			minChars: 2, // minimum query length
			debounce: 250, // ms to wait after the last keystroke
			perTypeLimit: 20, // rows fetched per content type
			maxResults: 100, // cap on the merged, ranked result set
			deep: true, // search inside components
			maxDepth: 3, // how many component levels to walk
			includeDrafts: true, // include unpublished entries
			queryTimeout: 4000, // per-content-type timeout in ms
			concurrency: 8, // content types queried in parallel
			excludedContentTypes: [], // e.g. ['api::log.log']
			excludedFields: [], // e.g. ['api::article.article:body']
			displayFields: {}, // extra fields in the result row, keyed by UID
		},
	},
};

Permissions

Two RBAC actions are registered and can be assigned per role:

• plugin::global-search.read — use the search page and the palette.
• plugin::global-search.settings — change the configuration.

Search results are additionally filtered by each user's Content Manager read permissions.

Performance notes

Searching runs live database LIKE queries — there is no index to maintain, but a query on a very large table is only as fast as your database makes it. The plugin keeps this bounded with a per-content-type row limit, a bounded column projection, a per-type timeout, a concurrency cap and a client-side debounce. For very large tables, consider a database index on the fields people actually search, or disable those content types in Settings.

Reported result counts reflect the capped set (perTypeLimit rows per type, maxResults overall) — a + in the UI means there were more matches than were fetched.

Limitations

• Dynamic zones and blocks fields are not searched: the query engine cannot filter polymorphic JSON columns.
• Relations, media and JSON fields are not searched.
• Only api:: content types are searched — not plugin content types such as users-permissions users or uploaded files.

Compatibility

Publishing and supply chain

Public releases are published from GitHub Actions with npm trusted publishing (OIDC). There is no long-lived 2FA-bypass granular access token. See npm install-time security and GAT bypass2fa deprecation.

License

MIT