Description:

Feature Description

Add functionality to connect with LLM models via an OpenAI-compatible API directly within the OASM Assistant settings. This allows users to configure custom Large Language Models (such as Local LLMs, Ollama, or other third-party providers) by providing the following parameters:

• API URL: The endpoint address for the LLM service.
• API Key: The authentication token required to access the API.
• Model Name: The specific model identifier to be used for requests.

Problem or Need

Currently, the OASM Assistant lacks a flexible way for users to define their own AI backend. Users need the ability to integrate with diverse AI infrastructures to ensure data privacy (via self-hosted models), optimize costs, or leverage specific models tailored for security analysis. Providing an OpenAI-compatible interface is the industry standard for such integrations.

Proposed Solution

In the OASM Assistant settings section, implement a configuration interface that includes:

1. Input fields for API URL, API Key, and Model Name.
2. A mechanism to save these credentials securely.
3. Integration logic within the Assistant to route prompts through the configured OpenAI-compatible provider instead of a hardcoded service.

Alternatives Considered

No response

Additional Context

This feature focuses specifically on the Assistant's settings to allow end-users to customize their "AI brain" without needing to modify global system or MCP server configurations.

Code of Conduct

• I agree to follow this project's Code of Conduct