Yes, indeed. But I went back to
/usr/bin/sailfish-qml harbour-datapass
which used to be there before editing.
The AI just replaced the line - (really ). I uploaded the updated package
Is there an important difference if I use the /usr/bin/invoker approach?
Meet sailorAI, a LLM Chat App supporting openAI and Gemini (Anthropic is untested). Chat with different models as long there is an API Key, Chat History etc. It is purely Claude driven (Using the Pro Version, there is not API access). No deps, API Calls are done via REST. Code is here: GitHub - conradhuebler/harbour-sailorAI: Native Sailfish LLM Chat App
And I udapted the datapass as well, I suppose both apps can go to the official store …
I took some time to give the repos some love and added CI RPM packages. Have only tested the packages via SDK.
I use sailorAI with ollama/gemma4-cloud, and tested also openai-api-endpoints. Anthropic was tested via the ollama endpoint, gemini was not tested recently. I am sure, I will extend sailorAI further.
Feature: Support for openai-api and ollama cloud endpoints, llm chat and AI based image description and and llm base image content translation. Including taking picture from within the app to have “translator app”.
Image model has to be choosen and is identified via api-tag. I used gemma4:cloud from ollama.