The v3 release has all the architectural components in place, but the UI and server concepts of a "workspace" aren't yet tied to the AI/OpenCode runner's session contexts. The OpenCode runner for v3.0.0 does not even have the ability to connect to CQL Studio Server's PostgreSQL database.
What should probably happen is that, in the IDE, the user should be required to select or create a workspace, as managed under Team -> Workspaces, prior to starting an OpenCode session. The runner would then read all the FHIR references affiliated with the workspace and add the metadata to the opencode session. Given that this includes any number of arbitrary FHIR resources, trying to put everything in the LLM context is a bad idea. However, it would be very useful for the LLM to have some sort of metadata of the data context and relevant identifiers used throughout the workspace, especially since there is no editor UI for general FHIR resources.
The v3 release has all the architectural components in place, but the UI and server concepts of a "workspace" aren't yet tied to the AI/OpenCode runner's session contexts. The OpenCode runner for v3.0.0 does not even have the ability to connect to CQL Studio Server's PostgreSQL database.
What should probably happen is that, in the IDE, the user should be required to select or create a workspace, as managed under Team -> Workspaces, prior to starting an OpenCode session. The runner would then read all the FHIR references affiliated with the workspace and add the metadata to the opencode session. Given that this includes any number of arbitrary FHIR resources, trying to put everything in the LLM context is a bad idea. However, it would be very useful for the LLM to have some sort of metadata of the data context and relevant identifiers used throughout the workspace, especially since there is no editor UI for general FHIR resources.