| 开发者 | alessandrolioce |
|---|---|
| 更新时间 | 2026年8月13日 22:55 |
| PHP版本: | 7.4 及以上 |
| WordPress版本: | 7.1 |
| 版权: | GPL-2.0+ |
| 版权网址: | 版权信息 |
render_block_core/query filter in block themes, and via the loop_start action in classic themes.creator-assistant-search folder to the /wp-content/plugins/ directory, or install the plugin directly from the WordPress plugin screen.No. Creator Assistant Search is an add-on that requires Creator Assistant - Hub to be installed and configured. The Hub provides the OpenAI and Qdrant connections that this plugin relies on.
By default, posts and pages are vectorized and included in search results. If you have Creator Assistant - Hub Pro, you can exclude specific post types from the Content tab in the Hub settings page.
The AI Overview appears when the plugin finds at least one relevant content chunk for the search term. If no relevant content is found in your Qdrant collection, the standard search results are shown without the overview panel.
This usually means your content has not been vectorized yet. Vectorization is handled entirely by Creator Assistant - Hub: new and updated posts are vectorized on save, and existing posts are picked up by the Hub's hourly WP-Cron batch, which requires WP-Cron to be running on your site. See the "How do I check whether the service is properly configured?" question in the Creator Assistant - Hub readme for the Site Health check and WP-Cron troubleshooting steps.
The overview is generated asynchronously via Server-Sent Events: the page loads with the standard search results immediately. The AI answer is then streamed progressively into the panel character by character (typically within a few seconds). If the answer for the same query was already cached, it appears almost instantly.
Yes. The generated answer is cached for one hour using WordPress transients. Repeated searches for the same query within that window return the cached result without any OpenAI API call.
Administrators can bypass the cache for a specific search by appending ?cas_no_cache=1 to the search URL (e.g. https://example.com/?s=shoes&cas_no_cache=1). This forces a fresh OpenAI call and overwrites the cached result. The bypass is only available to users with the manage_options capability and has no effect for regular visitors.
No. Creator Assistant Search is entirely non-destructive. It injects the AI Overview panel above search results using standard WordPress hooks (render_block_core/query for block themes, loop_start for classic themes). No theme files are ever touched. Your posts, pages, and all their content remain exactly as they are.
Search queries are never stored in plain text in your database. The plugin uses an anonymous hash of the query as a transient cache key for the AI response. The cached answer expires automatically after one hour and is then deleted. No query logs, no user identifiers, no permanent records. The search term itself is sent to OpenAI to generate the AI answer (subject to OpenAI's privacy policy), but it is not stored persistently by this plugin.
No. The AI Overview is an additional panel displayed above your standard search results. It does not remove, reorder, or replace any of them. If no relevant content is found in your Qdrant collection, the standard results are shown without any panel at all.
Yes, in two ways:
creator-assistant-search__ai-overview that you can override in your theme stylesheet. The animated border and typed-text effect are provided by Creator Assistant - Hub's shared assets.render_block_core/query) and classic themes (loop_start).creator_assistant_search_query_no_answer (action, fired when the AI returns no usable answer), creator_assistant_search_sources (filter, customise the sources list), creator_assistant_search_script_data (filter, customise JS data), creator_assistant_search_get_prompt_context (filter, access prompt context for other add-ons).