claude-code - 💡(How to fix) Fix Escape key does not interrupt Claude when deep in tool chain (Remote Control / SSH sessions) [1 comments, 2 participants]

TL;DR

The issue can be mitigated by investigating the handling of interrupt signals within the Claude Code session, potentially requiring adjustments to how Escape key presses are captured and processed during tool chain execution.

Guidance

• Investigate how Claude Code handles interrupt signals (e.g., SIGINT) when the Escape key is pressed, to understand why the current implementation does not interrupt the tool chain as expected.
• Review the documentation or source code of Claude Code to see if there are any specific configurations or settings that need to be adjusted to enable proper interrupt handling during remote sessions.
• Consider testing the behavior with different types of tool chains or individual tool calls to isolate if the issue is specific to certain commands or the combination of commands.
• Look into whether there are any known issues or open tickets related to interrupt handling in Claude Code's issue tracker that might provide insight or a temporary workaround.

Example

No specific code example can be provided without more details on the internal implementation of Claude Code or its handling of keyboard interrupts.

Notes

The solution may depend on the specifics of how Claude Code is implemented, especially regarding its use of subprocesses for tool calls and how it captures and handles keyboard input during execution. The issue might also be related to the Remote Control environment (claude.ai) and its interaction with the local macOS environment.

Recommendation

Apply a workaround, as the root cause seems to be related to the handling of interrupts within Claude Code, and without a clear indication of a fixed version that addresses this issue, adjusting the current setup or waiting for an update seems prudent.