How Chat Requests Work Internally (REST & Permissions)

1. REST Route for Chat #

From the docs:

• Namespace: davix-ai-engine/v1
• Route: /chat
• Registered in register_rest_routes() of Davix_Chatbot.

This endpoint is responsible for receiving user messages and returning AI responses.

2. Permission Check #

Permission callback: chat_permission_check()

• Validates:
  • REST nonces (security).
  • Login status, especially when login_required is ON.
• If checks fail, the request is denied.

3. Handling a Chat Request #

Handler: handle_chat_request()

It performs the following steps:

1. Read Input:
   • Extracts the user’s message from the request.
   • Loads Chatbot options (name, login requirements, site search toggles, etc.).
2. Optional Site Search:
   • If site search is enabled (enable_site_search, search_posts, search_pages, search_products), the handler may perform WordPress queries to gather contextual content.
3. Build OpenAI Request:
   • Uses:
     • openai_api_key
     • openai_model
     • system_prompt
     • temperature
   • Attaches any site-search context to the prompt when configured.
4. Call OpenAI API:
   • Sends a chat completion request using your configured model and parameters.
5. Return Response:
   • Formats the AI response and returns it as REST JSON.

image IMAGE PLACEHOLDER: Screenshot of Debug Logs (chat channel) showing sample REST request and OpenAI response lines. IMAGE image

4. Free vs Pro Behavior in Handler #

The same handler is used in Free and Pro, but:

• In Free, model is effectively fixed to gpt-4o-mini (no model switch).
• In Pro, additional options (model select, human colors, extended search scopes) are recognized and affect context and output.

5. Debugging Chat Requests #

You can use the Debug Logs module:

• Turn on debug_chatbot_enabled in Davix AI Engine → Debug Logs.
• This logs chat-related events, such as REST calls, to a log file in debug/logs/.

This is further detailed in the Debug Logs category, but it is important for understanding Chatbot behavior.