Chat Participant

The @ubcode chat participant is a GitHub Copilot Chat integration that gives you an AI assistant with deep knowledge of both Sphinx-Needs and the ubCode extension. Because it has direct access to the ubCode and Sphinx-Needs documentation, it can answer questions about anything in either tool or do changes based on your request.

Typical queries you might have:

• “How do I configure sphinx-needs?”

• “Please move the sphinx-needs configuration from my conf.py into a ubproject.toml”

• “Please migrate to the newest sphinx-needs version”

• “What is the syntax for linking requirements to test cases?”

• “How do I filter needs in a needtable?”

Example of the chat participant in action, called from the VS Code Chat panel.

How to Use

Invoke the assistant by typing @ubcode in the GitHub Copilot Chat panel, followed by your question:

@ubcode How do I create a custom need type with extra options?

The participant retrieves the relevant documentation pages and generates a focused answer in the context of your question. No project needs to be open for general documentation questions, but when a workspace is active the participant also takes your project context into account.

Important

If a follow-up question is still about Sphinx-Needs or ubCode, prefix it with @ubcode again so the participant continues to use the documentation context rather than falling back to general Copilot behaviour.

Configuration

The following VS Code settings control the chat participant’s behaviour.

ubcode.chatParticipant.llmselection

The Copilot model used for the documentation retrieval phase, which is the internal step where the participant selects which documentation pages to include in its context. The model selected in the chat window is always used for generating the answer itself. Leave this empty (the default) to also use the chat window model for retrieval, or set a specific model identifier (for example gpt-4o) to pin the retrieval step to a different model.

The easiest way to change this is via the Select Retrieval Model command in the Command Palette, which opens a quick-pick of available models and writes the setting for you. Alternatively, you can edit it manually via File ‣ Preferences ‣ Settings (search for ubcode chatParticipant) or directly in settings.json.

Commands

The following commands are available in the Command Palette (Ctrl+Shift+P / Cmd+Shift+P) and are prefixed with ubCode:

• Clear Documentation Cache Forces the participant to re-fetch the ubCode and Sphinx-Needs documentation on the next request. Useful if you notice stale or missing information in answers.

• Select Retrieval Model Opens a quick-pick to change the model used for the documentation retrieval phase without editing settings manually.