Yes, this is a known issue with the current GitHub Copilot plugin (0.13.0) in Eclipse 2025-09 when using models like Claude Opus 4.6.

What’s happening is the run_in_terminal tool is being triggered by Copilot Chat to search your workspace, but Eclipse is not properly returning the terminal output back to the chat session. Because of that, the tool call never resolves, and the UI gets stuck on the “Running run_in_terminal tool” spinner.

This appears to be an integration bug between:

GitHub Copilot for Eclipse (0.13.0)

Eclipse 4.37.0

Tool-based model execution (Claude Opus 4.6)

Workarounds that currently help:

Cancel and restart the chat session
Close the Chat view completely and reopen it. In some cases, restarting Eclipse fully clears the stuck tool state.

Switch models
Try using a different model (e.g., GPT-based models if available). The issue seems more frequent when tool execution is involved with Claude models.

Disable terminal tool usage (if configurable)
If there’s a setting to reduce or disable automatic terminal/tool usage, turn it off so the model doesn’t attempt to call run_in_terminal.

Manually provide context
Instead of letting Chat search for files, paste the relevant file contents directly into the chat. That avoids triggering the tool.

Update plugin
Check for a newer version of the GitHub Copilot Eclipse plugin. These tool-execution bugs are actively being fixed.

If none of the above works, reporting this through the GitHub Copilot feedback channel (Help → Report Issue in Eclipse) with logs attached would help the team reproduce it.

Yes, this appears to be a known issue related to GitHub Copilot Chat in Eclipse when the assistant attempts to use the run_in_terminal tool and the result is not properly returned to the chat session.

What seems to be happening is:

-> Copilot attempts to search for files.

-> It falls back to using the terminal.

-> Eclipse opens the Terminal view and runs a command.

-> The output does not get passed back into the chat.

The chat remains stuck showing “Running run_in_terminal tool” with an indefinite progress indicator.

This usually indicates a tool execution bridge problem between Eclipse, the Copilot plugin, and the backend model.

Possible causes include:

1. Tool execution bridge failure between the IDE and Copilot.

2. Workspace indexing issues that cause Copilot to rely on terminal search.

3. Model-specific instability (some Claude variants have shown inconsistent tool handling in IDE integrations).

4. A plugin version compatibility issue with Eclipse 2025-09.

Workarounds you can try:

1. Switch models (if possible).
   Try using a GPT-based model instead of Claude Opus 4.6 to see if tool execution behaves more reliably.

2. Prevent terminal usage.
   Explicitly instruct Copilot not to use the terminal, for example:
   “Do not use the terminal. Only search using workspace context.”
   This can stop it from triggering the run_in_terminal tool.

3. Restart the session.
   Close the chat panel, restart Eclipse, and open a new Copilot chat session. Once the tool gets stuck, there is usually no way to recover within that session.

4. Reinstall the Copilot plugin.
   Uninstall Copilot, restart Eclipse, reinstall the latest version, and re-authenticate with GitHub.

5. Re-index the workspace.
   Make sure the project is fully indexed and that files are not excluded from the workspace.

If the chat is already stuck showing “Running run_in_terminal tool,” you will need to close that chat and start a new one, as there is currently no way to interrupt or recover the stuck tool execution from within the same session.

This behavior has been reported in various IDE integrations and is typically related to tool response handling rather than a user configuration problem.