[BUG] Spurious tool-use rejections and background tasks silently marked "stopped" during long sessions

Status Open
Reported on v2.1.212
Maintainer reply None cached
Activity 0 comments · opened Jul 17, 2026

Preflight Checklist

  • [x] I have searched existing issues and this hasn't been reported yet
  • [x] This is a single bug report (please file separate reports for different bugs)
  • [x] I am using the latest version of Claude Code

What's Wrong?

Environment: Windows 11, Claude Code CLI, PowerShell + Bash tools used interchangeably in the same session. Long-running session (multiple hours, dozens of tool calls) working on a full-stack app (FastAPI backend + Vite/React frontend).

Issue 1: Tool-use calls occasionally rejected without user intent

Multiple times during the session, a proposed tool call (a file Edit, a Bash/PowerShell command) came back as rejected/interrupted even though I hadn't intended to deny it. Each time I confirmed to the assistant that I wanted the action to proceed and it was retried successfully with no other side effects. One instance was explicitly caused by trying to review the diff before clicking approve; the others gave no indication of cause. This happened at least 3 times in one session.

Issue 2: Background dev-server tasks marked "stopped" between turns with no completion record

I had the assistant start long-running background processes (uv run uvicorn ... for the backend, npm run dev for the frontend) multiple times over the session. Repeatedly — at least 4 separate occasions — both background tasks showed up in a later turn as:

"X background shell command task(s) from the previous session have no completion record. They may have been stopped (via the UI, Monitor timeout, or agent teardown — these leave no transcript marker), or they may have been running when the previous Claude Code process exited."

This required restarting both servers each time before continuing verification work. It's unclear whether this is expected behavior for background tasks that outlive some session/turn boundary, but it happened often enough within a single continuous session (not across separate app launches) that it looked like a recurring issue rather than an edge case.

Impact: Neither issue caused data loss or incorrect changes — the assistant caught both cases, confirmed with me, and recovered — but they added friction and repeated restart cycles throughout the session.

What Should Happen?

I should be able to accept Claude's suggestion and keep on coding,

Error Messages/Logs

Steps to Reproduce

It's randomly happening. I don'tknow how to reproduce it reliably.

Claude Model

Sonnet (default)

Is this a regression?

I don't know

Last Working Version

_No response_

Claude Code Version

2.1.212

Platform

Anthropic API

Operating System

Windows

Terminal/Shell

VS Code integrated terminal

Additional Information

_No response_

View original on GitHub ↗