Summary

In older versions of the Zed editor, the /diagnostic command was incredibly valuable, allowing users to inject errors reported by the project's language server directly into the Large Language Model (LLM) context within the AI Assist Panel. This feature was crucial for quickly getting an overview of current issues, warnings, and errors, enabling more efficient debugging and understanding of project health. Similarly, the /terminal command was also a key asset.

Currently, these powerful commands (/diagnostic and /terminal) are accessible within the "New Text Thread" section of the AI Assist Panel. However, they are notably absent from the "New Thread" functionality, which appears to be designed for more contextual, cursor-aware AI interactions.

The absence of these commands in "New Thread" significantly impacts my workflow, as it limits the ability to leverage the LLM effectively for dynamic project diagnostics and terminal interactions directly within the relevant code context.

I would like to request the availability of the @diagnostic and @terminal commands (or their equivalent features) within the "New Thread" functionality of the AI Assist Panel. This would enable users to feed current language server diagnostics (errors, warnings, hints) and terminal output into the LLM's context directly from the active "New Thread."

Ideally, this feature, when implemented in "New Thread," would:

• Be easily accessible, perhaps via a command palette entry or a dedicated UI element within the "New Thread" interface.
• Allow for filtering of diagnostic types (e.g., only errors, or all diagnostics).
• Provide a clear way to see what diagnostic/terminal information has been sent to the LLM.

Zed Version and System Specs

Zed: v0.189.0 (Zed Dev fc8702a)
OS: Windows 10.0.26100
Memory: 15.7 GiB
Architecture: x86_64
GPU: Intel(R) Iris(R) Xe Graphics || Intel Corporation || 101.6790