Content fetching

Content fetching is not publishing. It lets Edisource query your existing archive and pull a full article when the Agent needs context.

You configure it on Settings → CMS Publishing, under Content Fetching. The publishing credentials above that section are a separate record.

If no archive is saved, the page shows No archive connected. Click Connect archive.

Edisource only reads from this API. It never creates, edits, or deletes content.

Choose where the archive is hosted

The editor asks Where is your archive hosted?

Type	What you enter
WordPress	Public site address. Edisource calls the site’s REST API
Ghost	Site address plus a read-only Content API key
Custom API	A request and response mapping for any JSON API

WordPress

Uses the public REST API. You only need the site address. /wp-json must be reachable from the internet.

Field	What to paste
Connection Name (Optional)	A label such as WordPress archive
Site Address	The public website, such as https://www.example.com

After you enter a valid address, the form shows the two requests Edisource will call: a search (?search=…) and a single-post fetch (?slug=…).

Ghost

Uses the Ghost Content API (read-only). Do not paste the Admin API key from Connect Ghost.

Field	What to paste
Connection Name (Optional)	A label such as Ghost archive
Site Address	The Ghost site, without /ghost at the end. Example: https://www.example.com or https://example.ghost.io
Content API Key	In Ghost go to Settings → Integrations → Add custom integration and copy the Content API key

The preview shows a title search and a slug fetch under /ghost/api/content/posts/.

Custom API

Use this when the archive is not public WordPress or Ghost. Describe two requests: one that searches, and one that returns the full text of a single article.

The editor opens in the Visual editor. Switch to JSON if you prefer to paste the object.

Put placeholders in query params or the body. Edisource replaces them at request time ({keyword}, {searchQuery}, {pageSize}, {postUrlValue}, and similar). Add the path of any API key or token to encryptedFields so it is stored encrypted — for example fetchSearchResultsEndpoint.headers.Authorization.

View Documentation on the form opens the field reference.

What the schema does

The configuration is one JSON object. It has a few top-level fields and two endpoints:

Field	Required	Purpose
friendlyName	Yes	A label your team will recognize
cmsType	Yes	Short type name: wordpress, ghost, custom, or similar
encryptedFields	No	Dot-paths of secrets to encrypt
fetchSearchResultsEndpoint	One of the two endpoints	Search the archive
fetchPostContentEndpoint	One of the two endpoints	Fetch the full body of one article

Configure at least one endpoint. Most newsrooms need both.

Placeholders

Put these tokens in queryParams or bodyTemplate. Edisource replaces them at request time.

Placeholder	Meaning
{searchQuery} or {keyword}	The search term
{pageSize}	Results per page
{pageNumber}	Current page
{pageOffset}	(pageNumber - 1) * pageSize
{dateFrom} / {dateTo}	Date range (ISO 8601)
{postUrlValue}	Canonical URL of one article, used on fetchPostContentEndpoint
{postId}	ID of one article, when the fetch URL uses /{postId}

Map each token in placeholderMappings so Edisource knows which query or body key it belongs to.

Response mapping

Tell Edisource where the data lives in your JSON, using dot paths.

Search (fetchSearchResultsEndpoint.responseMapping)

Key	Required	Example path
resultsArrayPath	Yes (empty means the response itself is the array)	data.articles or results
itemUrlPath	Yes	link or attributes.url
itemTitlePath	Yes	title or headline.main
itemSummaryPath	No	summary
itemDatePath	No	pubDate
itemAuthorPath	No	author.name
itemImageUrlPath	No	image.url

Single post (fetchPostContentEndpoint.responseMapping)

Key	Required	Example path
contentBodyPath	Yes	data.post.html or content.body

Example schema

A REST search plus a single-post fetch. Replace the URLs and mark the Authorization header for encryption.

{
  "friendlyName": "Archive search — production",
  "cmsType": "custom",
  "encryptedFields": [
    "fetchSearchResultsEndpoint.headers.Authorization"
  ],
  "fetchSearchResultsEndpoint": {
    "url": "https://api.example.com/search",
    "method": "GET",
    "headers": {
      "Authorization": "Bearer YOUR_API_KEY",
      "Accept": "application/json"
    },
    "queryParams": {
      "q": "{searchQuery}",
      "limit": "{pageSize}"
    },
    "placeholderMappings": {
      "query": "{searchQuery}",
      "page_size": "{pageSize}"
    },
    "responseMapping": {
      "resultsArrayPath": "results",
      "itemUrlPath": "link",
      "itemTitlePath": "title",
      "itemSummaryPath": "summary"
    }
  },
  "fetchPostContentEndpoint": {
    "url": "https://api.example.com/posts/{postId}",
    "method": "GET",
    "queryParams": {},
    "placeholderMappings": {
      "url": "{postUrlValue}"
    },
    "responseMapping": {
      "contentBodyPath": "content.body"
    }
  }
}

GraphQL works the same way: put the query in bodyTemplate, use {searchQuery} inside it, and map the response paths to the GraphQL envelope.

Encryption

Any secret in a custom schema must appear in encryptedFields as the full path to that value. Ghost Content API keys are encrypted automatically. After save, Edisource encrypts those fields. Do not put a live token in a ticket or a shared doc.

When it is used

Discover feeds and studio tools can search the connected archive when Access to search your CMS is enabled for that feed or tool. If search returns nothing, check the site address (WordPress / Ghost), the placeholder mapping, the results array path, and that the endpoint is reachable from the internet.