[Bug] Insufficient Debugging Visibility for Subagent Workflow Failures

Bug Description
Lack of introspection into subagent failures. I was asking a subagent to do work and it asked for a zany persmission. I said no and then asked why it needed it and the main agent was as clueless as me. There appears to be no way to post-hoc see into what the agent was doing; once I stopped it we lost any debuggability

Environment Info

• Platform: linux
• Terminal: gnome-terminal
• Version: 2.0.8
• Feedback ID: 9de2ef59-ab38-4ad4-bfb8-bb667edf6c08

Errors

[{"error":"Error: Request was aborted.\n    at xP.makeRequest (file:///home/matthew/.local/share/nvm/v22.17.1/lib/node_modules/@anthropic-ai/claude-code/cli.js:1025:3864)\n    at runNextTicks (node:internal/process/task_queues:65:5)\n    at process.processTimers (node:internal/timers:520:9)","timestamp":"2025-10-05T17:58:22.826Z"}]