LLM / AI

Configure OpenAI-compatible providers, prompts, and inspect LLM audit events.

Centralized Defaults

Global and per-system defaults used by AI features (provider/model). Models are suggestions only.

Stored as llm.systemDefaults.

Stored as llm.providerModels. Suggestions only.

Providers

OpenAI-compatible endpoints used by prompts.

Key Label Preset Base URL Enabled

Prompts

Named templates that call configured providers/models.

Key Label Provider Model Enabled

Test Prompt

Run a configured prompt with sample variables and see the response.


              

            

          


          
          

            

              

                
LLM Audit

                
Recent LLM completions (action = llm.completion).

              

            

            

              

                

                  
                  
                

                

                  
                  
                

                

                  
                  
                

              

              

                

                  
                  
                

                

                  
                  Page 1
                  
                  
                

              

              

                

                  
                    

                      Time
                      Prompt
                      Provider
                      Model
                      Status
                      Tokens
                    

                  
                  
                

              

            

          

        


        
        

          

            

              
Cost Tracking

              
Manual search over successful LLM completions (tokens + USD cost if available).

            

          

          

            

              

                
                
              

              

                
                
              

              

                
                
              

              

                
                
              

            

            

              

                
                Page 1
                
              

              

                
              

            

            

              

                
                  

                    Time
                    Prompt
                    Provider
                    Model
                    Tokens (in/out/total)
                    Cost (USD)
                    Cost source
                  

                
                
              

            

          

        


        

          
Storage model

          
Providers and prompts are stored in GlobalSetting rows as JSON under keys llm.providers and llm.prompts. Use the internal service:

          
const superbackend = require('@intranefr/superbackend');

const result = await superbackend.services.llm.call('tell_joke', {
  theme: 'universe',
}, {
  temperature: 0.8,
});

console.log(result.content);