WordPress

Reads a WordPress install over its REST API. Structured, faster and cleaner than crawling the rendered site — and it can reach drafts and private posts that a crawl never sees.

Shape: config-only for public content (credentials are optional and only widen reach). Provider id: wordpress.

Look at the site first

wordpress-discover reports the site's canonical URL, its timezone, and every post type WordPress exposes over REST:

divinci connectors wordpress-discover https://example.com

Tip

A custom post type registered without show_in_rest: true is invisible to the REST API entirely — and that is the most common reason a sync completes successfully and ingests nothing. If your CPT is missing from this list, it is a one-line fix on the WordPress side, not a Divinci problem.

Connect it

# Public content — posts and pages by default
divinci connectors wordpress https://example.com --vector <vectorId>

# Named post types (use the rest_base from wordpress-discover, not the type name)
divinci connectors wordpress https://example.com \
  --vector <vectorId> --post-types posts,pages,protocols

# Daily re-sync, swapping through a draft vector
divinci connectors wordpress https://example.com \
  --vector <vectorId> --recurring --swap

Options

| Flag | What it does | | --- | --- | | --vector <id> | Target RAG vector | | --post-types <list> | rest_base values to ingest (default posts,pages) | | --limit <n> | Max items per post type | | --modified-after <iso> | Only items modified at/after this ISO-8601 instant | | --category-slugs <list> | Restrict to these category slugs (posts only) | | --exclude-slugs <list> | Skip these slugs | | --recurring, --cadence <minutes> | Recurring sync (15-minute floor, default 1440) | | --swap | Sync into a draft vector, swap the live release on success | | --username <user> | WordPress username, for non-public content | | --include-non-published | Also ingest drafts and private posts |

Caution

If the target vector is attached to a published release, pass --swap, or the live-vector guard rejects every chunk write while the sync still reports success.

Non-public content

Drafts and private posts require a WordPress Application Password (WP 5.6+). Pass it via the environment rather than a flag — anything in argv is readable by any process on the machine via ps:

export DIVINCI_WP_APP_PASSWORD='xxxx xxxx xxxx xxxx xxxx xxxx'
divinci connectors wordpress https://example.com \
  --vector <vectorId> --username editor --include-non-published

Caution

--include-non-published puts unfinished and deliberately-restricted content in front of your assistant's users. It ingests publish, draft and private — never future (scheduled) or pending, which are embargoed.

If the site redirects (http→https, www→apex), connect using the resolvedSiteUrl that wordpress-discover reports: an Application Password is withheld across a cross-origin hop, so an authenticated connector only works against the canonical origin.

Sync and re-run

WordPress is one of the two providers that can be re-run headlessly:

divinci connectors sync <connectionId>
divinci connectors sync-status <connectionId>

See also

• Managing connections
• Connect an E-commerce Store