Killed: 9 (SIGKILL / OOM killer) — page 2
The kernel terminated Claude Code outright, usually for memory pressure.
368 issues · 73 open · 295 resolved (80%) · first seen Feb 25, 2025
First response · local
What to do now
Most likely: The local Claude Code process crashed while parsing or executing.
- Record the Claude Code version and the first stack frame before restarting.
- Try a fresh session with the same minimal input to distinguish corrupted state from a repeatable crash.
- Preserve logs and compare reports from the same version before downgrading.
These are conservative triage steps, not an official Anthropic fix. Use the issue and workaround evidence below before making a destructive configuration change.
Is this getting better or worse?
This class of problem is holding steady. 122 new reports in the last 90 days vs 106 in the 90 before — +15%. The open backlog peaked at 105 in 2026-08 and sits at 73 today.
Derived by counting each matching issue from its creation date until it closed. First-party
computation over the anthropics/claude-code tracker — the trend is not published
anywhere upstream.
Which builds report this
252 of these reports name the Claude Code build they were running, spanning 129 releases. Heaviest reporters:
- Claude Code v2.1.220
- Claude Code v2.1.233
- Claude Code v2.1.205
- Claude Code v2.1.246
- Claude Code v2.1.231
- Claude Code v2.1.226
- Claude Code v2.1.170
- Claude Code v2.1.160
Report volume, not a defect count: an issue can mention a build without that build having caused the failure. Versions come from the version string in the bug-report template, matched against real published tags.
Do these get fixed?
Median time to close is 29 days across 295 closures. Of the 295 closures with a recorded reason, 15% were closed as completed and 252 as not-planned or duplicate. The most recent completed fix landed in 2026-10.
Workarounds reported by the community
183 of these issues post a workaround someone says worked. The highest-engagement ones:
- Claude.ai MCP servers auto-injected into Claude Code without opt-in — causes OOM crashes on resource-constrained systems
- …r's Claude.ai account and load them into every CLI session. Workaround Manually editing to set: However, this is a cached flag and may be overwritten on next update/startup. Expected Behavior 1. This should be opt-in, no…
Found in the description of #20412 · resolved - [BUG] Critical memory regression in 2.1.27 - OOM crash on simple input
- …toring during 2.1.27 session 2.1.25 after rollback (stable) Workaround Roll back to 2.1.25: Environment Info - Platform: Linux (Ubuntu 24.04, kernel 6.8.0-90-generic) - Architecture: x86 64 (QEMU/KVM VM) - RAM: 8GB + 4GB…
Found in the description of #22042 · resolved - Claude Code Memory Leak - Process Grows to 120+ GB RAM and Gets OOM Killed
- …d data structures (conversation history, file caches, etc.) Workaround Currently, the only workaround is to periodically restart Claude Code before memory exhaustion occurs, which is highly disruptive to workflow. --- Pr…
Found in the description of #4953 · still open - Critical Memory Leak: Claude Code Consumed 129GB RAM and Caused System Freeze
- …there's unbounded caching 5. Review event listener cleanup Workaround Currently monitoring with: Plan to restart Claude Code when memory usage exceeds safe threshold.
Found in the description of #11315 · still open - [BUG] Memory leak in long-running idle Claude Code sessions
- Not a fix for the leak, but a workaround that helped me live with it: unload the idle sessions and resume them later. I packaged that as an open-source plugin, It shows how much memory each session's process tree holds,…
Found in the thread of #18859 · still open
Excerpts are quoted from the linked issue so you can judge relevance before opening it. Threads that only ask for a workaround, or report that there is none, are excluded. These are community suggestions, not official guidance from Anthropic — read the full thread before applying one.
Matching issues 101–200
Ranked by community engagement (reactions weighted over comments). Listing the top 300 of 368 matches; the statistics above cover all 368.
- OOM crash corrupts session data: custom names lost and sessions missing from resume
- [BUG] SIGKILL crashes 5-10x daily since Opus 4.6 (v24.13.0)
- Memory spike: 3 processes peaked at 12.7 GB combined on 18 GB machine (v2.1.34, macOS native)
- [Bug] GC death spiral causes process freeze with unresponsive Ctrl+C during long sessions
- [BUG] New Memory Leak .
- Desktop app SIGKILLs its own Claude Code workers on a healthy Mac, then blames endpoint security (gates on raw free pages, not memory pressure)
- Per-tool-call helper process (claude.exe) leaks to 11.6GB anon RSS and is OOM-killed during a slow sandboxed Bash command
- [BUG] Desktop app (macOS): scheduled sessions with background tasks leak harness processes indefinitely (task-notification race, follow-up to #39632)
- Runaway anonymous-memory growth (~14.5 GiB in ~5 min) kills the CLI — reproduces on BOTH native (Bun/JSC) and npm (Node/V8) installs
- [BUG] macOS: foreground Bash completion leaves background child orphaned under PID 1 (2.1.220)
- [BUG] Claude Desktop (macOS): instances quit silently — clean beforeQuit, no crash report, not memory, active window always survives
- MCP server processes not terminated when Claude Code session exits
- Headless --resume/--fork-session reifies the entire session transcript in memory: 9-14x file size within seconds; 12.4 GB balloon OOM-killed a 16 GB host
- [BUG] Bundled claude-code CLI binary is unsigned — fails Gatekeeper on managed Mac, breaks Chat/Cowork/Code with "crashed" error
- stdio MCP servers spawned via launcher commands (bun run, npx) leave orphaned grandchildren after CLI exit — consider killing the process group
- Headless -p session leaks to 10-15GB RSS while idle on long-running background Bash tasks (v2.1.205, Linux)
- [BUG] tui: "fullscreen" leaks OSC11/DA1 terminal query-replies to stdout — corrupts piped output and pager navigation
- Deeply-nested subagent fan-out causes unbounded memory growth → host-level OOM
- [BUG] Sandbox recursively enumerates workspace into nested node_modules → unbounded memory → OOM (still repros on 2.1.195; cf #27863)
- [Windows] Auto-updater freezes when child processes (git, node, python) are still running
- Agent teams: teammate never exits on shutdown_request; pane never closed (killPane never called) — v2.1.185
- [BUG] Desktop local agent mode: spawn_task task sessions busy-loop at ~100% CPU and are never reaped (survive 44h+) — macOS 2.1.170
- [BUG] Desktop Code tab: /stats scan materializes entire multi-GB transcripts (incl. subagents/) → browser-process V8 heap OOM kills app; CLI handles same sessions fine
- [Behavioral] embedded-ripgrep V8-heap unbounded on file-listing (GH#54394 second spawn route)
- [BUG] claude rm/stop/reap SIGKILLs background session tree without SIGTERM grace, orphaning git index.lock and similar
- [Enhancement] Reap MCP child processes on parent claude exit — portable-workspace external drives get corrupted by orphan handles
- [BUG] Claude Code escapes systemd transient scope at startup, defeating user-imposed memory caps
- Misleading 'Blocked by endpoint security' popup on personal Mac with no endpoint-security software
- [BUG] Claude Desktop 1.5354.0: stdio MCP servers double-spawn from independent CCD/LAM coordinator registries
- Memory leak: RSS grows to ~6 GB on ARM64 in 2.1.121 (regression similar to #28763)
- Extended thinking burns unlimited tokens on 'continue' with large context — no output, no timeout, requires SIGKILL
- postinstall strips Mach-O code signature on darwin-arm64 → macOS SIGKILLs claude on launch
- [BUG] Channel plugin bun processes orphan on session exit, accumulate indefinitely
- app-server-broker.mjs processes leak when Claude Code session ends abnormally
- claude -p with --output-format text hangs after completing — MCP servers not cleaned up
- Bash-tool snapshot-source wrappers accumulate during subagent dispatch, triggering process-cascade OOM
- [BUG] Unbounded memory usage in LXD VMs where tool calling is used (CLI + VSCode Extension)
- Task output files in tmpfs grow unbounded, can consume all system RAM
- [BUG] Stop button fails to halt runaway agent — requires 5+ clicks and stop messages
- [Bug] statusLine command: zombie subprocess accumulation exhausts process table
- Claude Code crashes with OOM (2.6GB+) — no memory self-monitoring or graceful degradation
- [BUG] Claude Code freezes when Terminal.app menu bar / toolbar is activated (macOS)
- [BUG] Out of memory--ubuntu
- Ctrl+Z sends SIGSTOP to own PID instead of SIGTSTP to process group, breaks job control in containers/sandboxes
- Subagent processes inherit unbounded V8 heap reservation, causing OOM on large-memory systems
- [BUG] Marketplace installation silently fails (SIGKILL) on enterprise machines due to EDR blocking --recurse-submodules
- [BUG] Memory leak: multiple Claude processes accumulate 11GB+ RSS on Ubuntu 25.10 (kernel 6.17)
- Frequent hangs when run from OpenClaw via background PTY
- [BUG] Claude Code installer OOM when run as root during Docker build
- Startup freeze and 100% CPU zombie processes when VS Code extension and CLI versions mismatch
- [BUG] Memory leak - Claude Code 2.1.2 grows to 6-7GB RAM in short conversations
- [BUG] Question Clarification Tool Makes Claude Code Stuck sometimes
- [BUG] Stdio MCP servers still connecting (or queued) at session end are never terminated and are left orphaned
- Agent performed hours of unauthorized work, ignored its core process rule after eight corrections, and reported unverified results as fact
- Bundled ugrep shadowing `grep` can exhaust system memory — froze a 16 GB machine twice
- [BUG] claude ignores SIGINT despite a registered handler — Ctrl+C dead during startup hang, only SIGKILL works (Bun runtime)
- [BUG] Background Bash task's process group is SIGKILLed mid-run when the command spawns a daemonizing CLI (cursor-agent) — and reported as completed (exit code 0)
- [BUG] Runaway memory growth in Claude Code child process causes full system freeze (Ubuntu 24.04, requires hard reboot)
- Hook timeout does not reclaim a hook blocked in a read syscall — SIGTERM ignored, process survives indefinitely
- Remote Control: supervisor exits 5s after last client detaches, so it can't be left available overnight
- [BUG] Headless CLI reserves memory from host RAM, leaks to 11-31 GB, and takes down the entire WSL2 VM (2.1.181 / 2.1.215 / 2.1.220)
- telegram channel plugin: orphaned server.ts processes accumulate unbounded, CPU-busy, immune to SIGTERM
- [BUG] Telegram channel plugin: starting a second session kills the running poller, leaving the channel permanently dead
- grep shim (embedded ugrep 7.5.0) allocates unbounded memory on double-bounded-repeat BRE patterns — can OOM-kill the host
- macOS: intermittent 2-minute git hangs with core.fsmonitor enabled (multi-worktree/concurrent sessions); timeout SIGTERM orphans index.lock and wedges the session
- Background session crash-loop on attach + ambiguous kill behavior on worker pool
- [BUG] Fable 5 = "Self-caged" : the model's own output (`fork`/`kill`/`orphan`/`reap`) trips its own classifier midstream — input rewording can't fix it
- Background-agent daemon self-SIGKILLs its process group (kill(0, SIGKILL)) every ~50s, killing FleetView agents — 2.1.195 & 2.1.196
- Claude Code goes silent when using VS Code Remote SSH Extension and Claude Code on Digital Ocean droplet
- telegram plugin: bun server.ts leaks at 100% CPU on every MCP disconnect
- [BUG] Workflow tool: runner subprocess leaks memory unbounded (~70 MB/s) → OOM-killed on long multi-agent runs
- /mcp reconnect after config-hash-stale spawns literal unexpanded ${CLAUDE_PLUGIN_ROOT} → -32000
- [BUG] Headless `--output-format stream-json` sessions spawned by desktop-app scheduled tasks (with Task subagent fan-out) fail to retire after completion
- Spawned child process (versioned binary) grew to 12.3 GB RSS and was OOM-killed
- [BUG] CLI self-aborts (SIGABRT) at ~6 GB RSS during long background Workflow runs — reproduced on 2.1.160 and 2.1.161 (WSL2)
- [DOCS] Background session docs omit SIGTERM-first teardown for shell subprocesses
- [BUG] Native memory leak (~1.5 GB/min RssAnon) in interactive mode — regression between 2.1.140 and 2.1.150
- [BUG] Switching focus between local Code sessions kills the unfocused session's CLI process (SIGTERM, exit 143) instead of leaving it running in the background
- `claude remote-control` daemon leaks idle `--print` child sessions (never reaped)
- MCP STDIO subprocess reaped + respawned mid-conversation, no shutdown signal to the server
- claude --print hangs on Sonnet 4-6 at default/high effort with large prompts
- Detached worktrees from subagent isolation accumulate silently with no lifecycle management
- [BUG] Claude Desktop scheduled-tasks runner leaves orphan CLI subprocesses
- Bun runtime SIGKILL's stdio MCP server process group after ~60s idle
- claude -p wedges (600s timeout) when invoked via Python subprocess.run with --model sonnet, --output-format json, and large prompt content (macOS arm64)
- VS Code extension does not reap orphan child claude processes
- /resume picker shows empty after process exit, regardless of /rename, real prompts, or synthesized live PID metadata
- [BUG] skill-creator run_loop.py: --results-dir stays empty when parent session compacts (no per-iteration flush + lost task handle)
- Session JSONL corrupted by non-atomic append when process killed mid-write — `--resume` fails with JSON parse error
- [Proposal] backgroundMaxElapsedSeconds — watchdog for Bash processes left by run_in_background (16 h silent zombie in auto mode)
- [BUG] SIGKILL on older CPU
- [BUG] Claude Code hangs mid-session after extended tool-use; prompt disappears, SIGTERM ignore
- [BUG] Bash tool sends SIGTERM instead of SIGINT when cancelling commands, corrupting build state
- Orphaned Claude Code process enters 100% CPU busy loop when parent terminal is closed
- [BUG] Stale MCP Session ID Cached Permanently
- Code tab unresponsive: Electron env vars (OPERON_SANDBOXED_NETWORK) break bundled CLI
- 2.1.97 binary gets SIGKILL'd on macOS 26.3 (Tahoe beta)
- Telegram plugin leaks orphan bun server.ts processes across session restarts; stuck ones pin a full CPU core indefinitely
- Telegram channel plugin: bun server.ts processes consume ~80% CPU when idle
- Orchestrator crashes on burst of simultaneous background agent completions