Cache Manager — Strapi 5 Plugin

A plug-and-play Strapi 5 plugin that gives editors manual control over cache invalidation directly from the admin UI. It abstracts away the cache implementation — Varnish, Redis proxy APIs, CDN purge APIs, or any HTTP-based cache system can be configured as a provider without touching plugin code.

---

Features

• Document Action — "Purge Cache" in the edit view dropdown and in list-view table row actions. Purges the cache entry for the specific content being edited.
• Bulk Action — "Purge Cache (N)" in the list-view bulk action toolbar. Deduplicates paths across all selected entries before purging, so /news is only called once even if 20 articles are selected.
• Settings Dashboard — Settings > Cache Manager shows all configured providers, their available endpoints, live connection statistics, per-provider "Purge" buttons, a "Purge All Caches" control, and a read-only view of the content type mapping configuration.
• Provider-agnostic — Communicates with any HTTP-accessible cache system. All connection details live in config/plugins.ts.
• Environment variable interpolation — Use ${VAR_NAME} in any URL or header value. Resolved at request time from process.env, so credentials are never hardcoded.
• Internationalized — Ships with English and German translations. Additional languages can be added by dropping a JSON file in admin/src/translations/.

---

How It Works

The Big Picture

Strapi Admin UI
  ├── Edit view → Document Action ("Purge Cache")   ─┐
  ├── List view → Bulk Action ("Purge Cache (N)")    ├──▶  Admin API (authenticated)  ──▶  Cache Service  ──▶  Provider HTTP endpoints
  └── Settings → Cache Manager Dashboard            ─┘         /cache-manager/*             (server)            (Varnish, CDN, ...)

When an editor triggers a purge:

1. The admin UI calls one of the plugin's server-side routes, authenticated via the Strapi admin JWT.
2. The server fetches the content entry from Strapi's document service to get its field values (slug, etc.).
3. The cache service resolves which frontend URL paths need purging, using the content type mapping defined in config.
4. For each resolved path, the service calls the configured provider endpoints over HTTP — for example, Varnish's /varnish-purge and /varnish-ban endpoints.
5. Results are aggregated and returned. The success field reflects whether at least one provider call succeeded.

Content Type → URL Path Resolution

The contentTypeMapping config tells the plugin how to translate a Strapi content type entry into frontend URL paths. For example:

api::post.post + { slug: "my-post" }
  → pathPattern "/blog/{slug}"  →  "/blog/my-post"
  → relatedPaths               →  "/blog", "/"
  → purge: ["/blog/my-post", "/blog", "/"]

The {fieldName} placeholder in pathPattern is replaced with the actual value of that field from the entry. Any top-level entry field can be used — {slug}, {locale}, {id}, etc.

Content types with purgeAllOnChange: true skip path resolution entirely and ban everything (useful for global content like headers, footers, and navigation menus).

Varnish Purge Strategy: Purge + Ban

For each path, the plugin calls both the purge and ban endpoints on providers that have them configured:

• purge (POST /varnish-purge with X-Purge-URL): Removes the exact cached object immediately. Fast and precise.
• ban (POST /varnish-ban with X-Ban-URL): Adds a ban expression to Varnish's ban list. Invalidates any object whose URL matches the pattern, including any grace-period or stale objects that purge might miss.

Together they provide thorough cache invalidation: purge handles the immediately cached object, ban handles any lingering stale variants.

---

Setup

Prerequisites

• Strapi 5.x
• Node 22+

Step 1 — Install

npm install strapi-plugin-cache-manager
# or
pnpm add strapi-plugin-cache-manager
# or
yarn add strapi-plugin-cache-manager

Step 2 — Configure the plugin

In config/plugins.ts:

export default ({ env }) => ({
  'cache-manager': {
    enabled: true,
    config: {
      providers: [
        {
          name: 'Varnish',
          type: 'http',
          endpoints: {
            purge: {
              url: '${VARNISH_URL}/varnish-purge',
              method: 'POST',
              headers: { 'X-Purge-Token': '${VARNISH_PURGE_TOKEN}' },
              pathParam: 'X-Purge-URL',
              pathLocation: 'header',
            },
            ban: {
              url: '${VARNISH_URL}/varnish-ban',
              method: 'POST',
              headers: { 'X-Purge-Token': '${VARNISH_PURGE_TOKEN}' },
              pathParam: 'X-Ban-URL',
              pathLocation: 'header',
            },
            purgeAll: {
              url: '${VARNISH_URL}/varnish-ban',
              method: 'POST',
              headers: {
                'X-Purge-Token': '${VARNISH_PURGE_TOKEN}',
                'X-Ban-URL': '.',
              },
            },
          },
        },
      ],
      contentTypeMapping: {
        'api::post.post': {
          pathPattern: '/blog/{slug}',
          relatedPaths: ['/blog', '/'],
        },
        'api::page.page': {
          pathPattern: '/{slug}',
        },
        'api::footer.footer': {
          purgeAllOnChange: true,
        },
        'api::tag.tag': {
          relatedPaths: ['/blog'],
        },
      },
    },
  },
});

Step 3 — Set environment variables

# .env
VARNISH_URL=http://localhost:6081       # local dev (Varnish exposed on port 6081)
VARNISH_PURGE_TOKEN=your-secret-token  # must match the token Varnish is configured with

In production (Docker), use VARNISH_URL=http://varnish (the container name).

Step 4 — Start Strapi

pnpm develop

Navigate to Settings > Cache Manager to verify providers are showing.

---

Configuration Reference

Provider Shape

{
  name: string;    // Display name shown in the dashboard
  type: string;    // Informational type tag (e.g. 'http')
  endpoints: {
    purge?: ProviderEndpoint;     // Called for each path on single/bulk purge
    ban?: ProviderEndpoint;       // Called for each path on single/bulk purge (Varnish ban list)
    purgeAll?: ProviderEndpoint;  // Called when "Purge All" is triggered
    stats?: ProviderEndpoint;     // Called on dashboard load to show statistics
  };
}

All endpoints are optional. A provider with only purgeAll is valid. A provider with only stats is valid.

Endpoint Shape

{
  url: string;                         // Base URL — supports ${VAR_NAME} interpolation
  method: 'GET' | 'POST' | 'PUT' | 'DELETE';
  headers?: Record<string, string>;    // Extra headers — supports ${VAR_NAME} interpolation
  params?: Record<string, string>;     // Static query parameters added to every request
  pathParam?: string;                  // Name of the param or header that receives the path value
  pathLocation?: 'query' | 'header';  // Where the path value is sent (default: 'query')
}

Path Delivery: pathParam and pathLocation

When purging a specific URL (e.g. /news/my-article), the plugin needs to tell the cache system which URL to purge. Different systems expect this in different places:

Via query parameter (pathLocation: 'query', the default):

GET http://my-cache/purge?path=/news/my-article
                    ^^^^
                    pathParam: 'path'

Via HTTP header (pathLocation: 'header'):

POST http://varnish/varnish-purge
X-Purge-URL: /news/my-article
^^^^^^^^^^^^^^^^^
pathParam: 'X-Purge-URL'

Content Type Mapping

contentTypeMapping: {
  'api::post.post': {
    pathPattern: '/blog/{slug}',     // {fieldName} replaced with entry field value
    relatedPaths: ['/blog', '/'],    // Additional paths purged alongside the entry path
  },
  'api::footer.footer': {
    purgeAllOnChange: true,          // Triggers purgeAll instead of path-based purge
  },
  'api::tag.tag': {
    relatedPaths: ['/blog'],         // No pathPattern — only related paths are purged
  },
}

Field	Type	Description
pathPattern	string	URL pattern — {fieldName} is replaced with entry[fieldName]
relatedPaths	string[]	Extra paths purged alongside the entry path (e.g. listing pages)
purgeAllOnChange	boolean	When true, triggers a full cache purge instead of specific paths

Field substitution examples:

pathPattern: '/blog/{slug}'          + entry.slug = 'my-post'    →  '/blog/my-post'
pathPattern: '/{locale}/blog/{slug}' + entry.locale = 'en', entry.slug = 'my-post'  →  '/en/blog/my-post'
pathPattern: '/products/{slug}'      + entry.slug = 'my-product' →  '/products/my-product'

Any top-level field on the Strapi entry can be used as a placeholder.

Environment Variable Interpolation

Use ${VAR_NAME} anywhere in a URL, header value, or param value:

url: '${VARNISH_URL}/varnish-purge'          // resolved at request time
headers: { 'X-Purge-Token': '${VARNISH_PURGE_TOKEN}' }

Resolution happens at request time (not at startup), so:

• Credentials are never exposed in the config or the database
• The admin UI never receives URLs or header values (only provider names and endpoint keys)
• A missing ${VAR_NAME} resolves to an empty string (does not crash Strapi)

---

Provider Configuration Examples

Varnish

Varnish accepts purge and ban via custom HTTP endpoints defined in default.vcl:

{
  name: 'Varnish',
  type: 'http',
  endpoints: {
    purge: {
      url: '${VARNISH_URL}/varnish-purge',
      method: 'POST',
      headers: { 'X-Purge-Token': '${VARNISH_PURGE_TOKEN}' },
      pathParam: 'X-Purge-URL',
      pathLocation: 'header',
    },
    ban: {
      url: '${VARNISH_URL}/varnish-ban',
      method: 'POST',
      headers: { 'X-Purge-Token': '${VARNISH_PURGE_TOKEN}' },
      pathParam: 'X-Ban-URL',
      pathLocation: 'header',
    },
    purgeAll: {
      url: '${VARNISH_URL}/varnish-ban',
      method: 'POST',
      headers: {
        'X-Purge-Token': '${VARNISH_PURGE_TOKEN}',
        'X-Ban-URL': '.',   // regex '.' matches every URL
      },
    },
  },
}

Required env vars: VARNISH_URL, VARNISH_PURGE_TOKEN

How it works: The VCL checks X-Purge-Token on every purge/ban request. If it matches, the operation proceeds; otherwise Varnish returns 403. The purgeAll endpoint bans everything by sending X-Ban-URL: . (regex that matches all URLs).

Redis (via HTTP proxy endpoint)

Redis has no native HTTP API, so you expose a small endpoint in your frontend app (Next.js, Astro, Express, etc.) that handles Redis operations and call it from the plugin.

Example frontend endpoint (/api/cache):

GET  /api/cache?action=clear-path&path=/blog/my-post  → deletes the Redis key for that path
GET  /api/cache?action=stats                           → returns key counts / memory usage
DELETE /api/cache                                      → flushes all cached keys

Plugin config:

{
  name: 'Redis Cache',
  type: 'http',
  endpoints: {
    purge: {
      url: '${FRONTEND_URL}/api/cache',
      method: 'GET',
      headers: { Authorization: 'Bearer ${CACHE_API_TOKEN}' },
      params: { action: 'clear-path' },
      pathParam: 'path',
      pathLocation: 'query',   // sends path as ?path=/blog/my-post
    },
    purgeAll: {
      url: '${FRONTEND_URL}/api/cache',
      method: 'DELETE',
      headers: { Authorization: 'Bearer ${CACHE_API_TOKEN}' },
    },
    stats: {
      url: '${FRONTEND_URL}/api/cache',
      method: 'GET',
      headers: { Authorization: 'Bearer ${CACHE_API_TOKEN}' },
      params: { action: 'stats' },
    },
  },
}

Required env vars: FRONTEND_URL, CACHE_API_TOKEN

How it works: The plugin calls your frontend's HTTP endpoint for each cache operation. The endpoint translates the request into the appropriate Redis command (DEL for a specific key, FLUSHDB for purge-all, etc.). The Authorization header prevents unauthorized cache clears.

CDN (Cloudflare example)

{
  name: 'Cloudflare',
  type: 'http',
  endpoints: {
    purgeAll: {
      url: 'https://api.cloudflare.com/client/v4/zones/${CF_ZONE_ID}/purge_cache',
      method: 'POST',
      headers: { 'Authorization': 'Bearer ${CF_API_TOKEN}' },
    },
  },
}

Note: Cloudflare's purge API requires a JSON body ({ "purge_everything": true }). The current plugin version does not send request bodies. Per-URL purge is not yet supported for APIs that require body payloads. See Extending the Plugin.

Multiple Providers

All providers are called for every operation:

providers: [
  { name: 'Varnish', type: 'http', endpoints: { purge: {...}, ban: {...}, purgeAll: {...} } },
  { name: 'Redis Cache', type: 'http', endpoints: { purge: {...}, purgeAll: {...}, stats: {...} } },
]

Purging one post calls Varnish's purge + ban for each path, and in parallel calls the Redis endpoint's purge for each path — both caches are invalidated in a single editor action.

---

Admin API Routes

All routes are served under Strapi's admin API prefix and require a valid admin JWT token (admin::isAuthenticatedAdmin policy). They are not accessible from the public Strapi REST/GraphQL API.

Method	Path	Body / Query	Description
GET	/cache-manager/providers	—	List providers (name, type, endpoint keys only — no URLs or credentials)
GET	/cache-manager/stats	?provider=Name (optional)	Fetch stats from providers that have a stats endpoint
GET	/cache-manager/content-type-mapping	—	Return the full content type mapping config
POST	/cache-manager/purge-entry	{ contentTypeUid, documentId }	Purge cache for one entry
POST	/cache-manager/purge-bulk	{ contentTypeUid, documentIds[] }	Purge cache for multiple entries (paths deduplicated)
POST	/cache-manager/purge-all	{ provider? } (optional)	Purge all caches (optionally scoped to one provider)

Response Shape

Every purge route returns:

{
  "success": true,
  "results": [
    {
      "provider": "Varnish",
      "endpoint": "purge",
      "success": true,
      "status": 200,
      "message": "purge succeeded on Varnish",
      "details": null
    },
    {
      "provider": "Varnish",
      "endpoint": "ban",
      "success": true,
      "status": 200,
      "message": "ban succeeded on Varnish",
      "details": null
    }
  ]
}

success at the top level is true when at least one result succeeded and there is at least one result. details contains the parsed JSON response body from the cache provider (if any).

Failure example (wrong token):

{
  "success": false,
  "results": [
    {
      "provider": "Varnish",
      "endpoint": "purge",
      "success": false,
      "status": 403,
      "message": "purge failed on Varnish: HTTP 403",
      "details": null
    }
  ]
}

Failure example (network error):

{
  "success": false,
  "results": [
    {
      "provider": "Varnish",
      "endpoint": "purge",
      "success": false,
      "message": "purge failed on Varnish: fetch failed",
      "details": null
    }
  ]
}

---

Plugin File Structure

cache-manager/
├── package.json                   strapi plugin package (kind: plugin)
├── README.md
├── ARCHITECTURE.md
│
├── admin/
│   ├── custom.d.ts                ambient type declarations
│   ├── tsconfig.json
│   ├── tsconfig.build.json
│   └── src/
│       ├── index.tsx              plugin registration: document action, bulk action, settings link
│       ├── pluginId.ts            PLUGIN_ID constant: 'cache-manager'
│       ├── api.ts                 authenticated fetch wrappers (getFetchClient)
│       ├── components/
│       │   ├── Initializer.tsx    marks plugin as ready after registration
│       │   └── PluginIcon.tsx     trash icon used in actions
│       ├── pages/
│       │   └── Settings.tsx       dashboard: providers table, stats, purge-all dialog
│       ├── translations/
│       │   ├── en.json            English strings
│       │   └── de.json            German strings
│       └── utils/
│           └── getTranslation.ts  prefixes translation keys with 'cache-manager.'
│
└── server/
    ├── tsconfig.json
    ├── tsconfig.build.json
    └── src/
        ├── index.ts               server entry: wires config, bootstrap, controllers, routes, services
        ├── register.ts            (empty — no custom fields)
        ├── bootstrap.ts           logs configured provider names on startup
        ├── destroy.ts             (empty — no cleanup needed)
        ├── config/
        │   └── index.ts           default config (providers: [], contentTypeMapping: {}) + validator
        ├── controllers/
        │   ├── index.ts
        │   └── cache-controller.ts  validates input, fetches entries, delegates to service
        ├── routes/
        │   ├── index.ts           declares route group as type 'admin'
        │   └── admin-api.ts       6 route definitions with isAuthenticatedAdmin policy
        ├── services/
        │   ├── index.ts
        │   └── cache-service.ts   core logic: env var resolution, path resolution, HTTP execution
        ├── content-types/
        │   └── index.ts           (empty — no custom content types)
        ├── middlewares/
        │   └── index.ts           (empty)
        └── policies/
            └── index.ts           (empty — uses built-in admin policy)

---

Translations

Translation keys follow this naming convention:

Key prefix	Where used
cache-manager.settings.*	Settings dashboard page
cache-manager.action.purge.*	Single-entry document action
cache-manager.action.purge-bulk.*	Bulk action

To add a new language, create admin/src/translations/{locale}.json following the same key structure as en.json. The registerTrads function in index.tsx discovers it automatically via dynamic import.

---

Extending the Plugin

Adding request bodies

Some APIs (e.g. Cloudflare) require a JSON body. To support this:

1. Add body?: Record<string, unknown> to the ProviderEndpoint interface in cache-service.ts.
2. In executeEndpoint, pass body: endpoint.body ? JSON.stringify(endpoint.body) : undefined to the fetch() call.
3. In plugins.ts, configure the endpoint with a body field.

Adding a custom provider type

All providers currently use the 'http' type. To add a provider that communicates differently (e.g. via native Redis commands):

1. Add a type guard in executeEndpoint that checks provider.type.
2. Implement the custom logic path for the new type.
3. The ProviderConfig.type field can remain freeform — it's informational only.

Lifecycle hook integration (automatic purge on publish)

The plugin is currently manual-only — editors trigger purges. To add automatic purging when content is published:

1. In server/src/bootstrap.ts, register lifecycle hooks using strapi.db.lifecycles.subscribe.
2. In the afterUpdate/afterPublish handlers, call strapi.plugin('cache-manager').service('cache-service').purgeEntry(uid, entry).
3. This would make cache invalidation transparent to editors.

---

Troubleshooting

Plugin doesn't appear in Settings:

• Check enabled: true and resolve: './src/plugins/cache-manager' in config/plugins.ts
• Build the plugin — it must be compiled before Strapi can load it: pnpm --filter cache-manager build
• Look for [cache-manager] messages in Strapi startup logs

"fetch failed" on purge:

• The cache provider isn't running or isn't reachable at the configured URL
• In local dev: start Docker (docker compose up -d) and verify your cache service is up
• Test the endpoint directly, e.g.: curl -X POST http://localhost:6081/varnish-purge -H "X-Purge-Token: your-token" -H "X-Purge-URL: /"

HTTP 403 on purge:

• VARNISH_PURGE_TOKEN in your .env doesn't match the token your cache provider expects
• Restart the container after changing the token: docker compose up -d --force-recreate varnish

"Invalid URL" on purge:

• VARNISH_URL (or whichever URL env var you're using) is not set in your .env
• The ${VAR_NAME} interpolation resolves to an empty string, making new URL('') throw
• Set it to the correct address, e.g. http://localhost:6081 (local dev) or the container name in production

Document action not showing in edit view:

• The action hides itself when documentId or document is null (e.g. when creating a new entry before first save)
• Check browser console for JavaScript errors
• Verify the content-manager plugin loads before cache-manager

Bulk action not showing:

• Select at least one entry in the list view — the action only appears in the bulk selection toolbar