Claude Code v2.1.201
Released Jul 3, 2026 (1 month ago).
⚠ Elevated report volume. This build drew 351 issue reports —
more than double the typical release (160 median). That can mean a regression, or simply
that many users happened to be on this build. Skim the reports below before deciding to upgrade.
351 issues reference this build
201 still open
150 resolved (43%)
"References this build" means the issue body mentions running 2.1.201 (from the
bug-report template). It reflects report volume, not a proven defect count — an issue may mention a
version without being caused by it. First-party data from the anthropics/claude-code tracker.
Top issues reported on v2.1.201
Ranked by community engagement (reactions and comments).
- Assistant text blocks silently dropped when followed by more thinking in the same turn — never rendered, missing from transcript JSONL (2.1.201, adaptive thinking)
- [BUG] Fable 5: mid-turn assistant text blocks intermittently delivered as summarized thinking blocks (turn appears silent)
- Background worker crash-loop on session start: "worker crashed (exit 1 — exit_with_message) — respawning…"
- [BUG] Burning through more tokens, while doing nothing different then before.
- Bundled claude-api skill injects entire 570KB SKILL.md (~210k tokens) into context on load
- [Bug] Fable 5 writes user-facing output into hidden thinking instead of a text block — same as #66112, fixed for Opus but still reproduces on Fable
- [BUG] Sessions containing /loop command are permanently hidden from /resume
- "Always allow" for Claude-in-Chrome site permissions is always persisted as duration:"once" — approved sites list stays empty, prompt repeats for every browser action
- WorktreeRemove hook never fires when a worktree is removed at session exit
- Claude Desktop: allow opening/resuming local sessions not present in the account's sidebar (multi-account support)
- /resume can't reopen sessions with live background tasks — forced detour through `claude agents`
- loop trying to login.
- [BUG] PermissionRequest hook 'allow' for ExitPlanMode is ignored since v2.1.199 — plan-approval chooser still blocks
- grep shell shim (claude-as-ugrep): unbounded memory growth on ERE bounded quantifiers {m,n} — OOMs the host
- [BUG] TUI renders garbled inside tmux since 2.1.200 (last-good 2.1.199) — only repaints on forced redraw
- Remote Control: no way to enable Claude in Chrome for spawned sessions
- Linux sandbox: recursive Read() deny globs expand to per-file bwrap binds → E2BIG on every command (incl. echo hello) when a denied directory holds many files
- Auto-compaction plateaus near ~75% context usage on Sonnet 5 (v2.1.201), causing repeated compact/work loop
- [BUG] MSIX Installation failure: AddPackage failed with HRESULT 0x80073CF9
- [BUG] PyCharm Terminal scroll
- [BUG] ScheduleWakeup in --print/headless mode is a silent no-op
- macOS TCC permission prompts show "2.1.201" as the app name, and grants reset on every version update
- [BUG] Subagent model override is dropped on resume/continuation — resumed turns silently fall back to the session model
- sonnet[1m] and fable[1m] model aliases do not apply 1M context window (only opus[1m] works)
- Vi mode: option for yank/put to use the system clipboard (like Vim's clipboard=unnamed)
- Background agent sessions (claude --bg/claude agents): fast session termination, worker crash-loop on attach, and lost background-task completion records
- [Bug] Claude Code fails to launch with EPERM error on macOS
- Fan-out research skills (deep-research) inherit the session model for all subagents — no model tiering — burning premium-model quota in a single run
- [FEATURE] Execute queued built-in slash commands (e.g. /compact) at end of turn instead of delivering them to the model as text
- [BUG] Claude Desktop window raises to front on mouse hover (LinuxMint/Cinnamon), unlike other Electron apps
Changelog
What's changed
- Claude Sonnet 5 sessions no longer use the mid-conversation system role for harness reminders