Claude Code v2.1.202
Released Jul 6, 2026 (1 month ago).
204 issues reference this build
145 still open
59 resolved (29%)
"References this build" means the issue body mentions running 2.1.202 (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 v2.1.202
Ranked by community engagement (reactions and comments).
- [BUG] Server-side experiment (`x-cc-atis`) silently removed Opus 4.8 thinking summaries, and the CLI silently self-updated even with `autoUpdates: false`. No notice, no opt-in, settings silently overridden.
- [FEATURE] Feature request: expose a "rename session" tool to the agent (ccd_session_mgmt)
- [BUG] Plan mode: cannot scroll to reach Claude's answers rendered above the plan box
- [BUG] Assistant text preceding an AskUserQuestion call in the same turn is never displayed (terminal CLI)
- [BUG] /compact and auto-compaction drop entire Available skills system-reminder; /reload-skills recovers with "no changes"
- [FEATURE] /workflows: attribute nested workflow() runs to the parent phase active at the call site
- [BUG] Persistent authentication loop across VS Code extension, CLI, and email login flow
- Mid-turn assistant text lost with interleaved-thinking models: destroyed on 2.1.202; persists but never renders on 2.1.203+ (default and transcript views)
- Desktop app's worktree pool reclaims/re-leases a directory while a session is still using it, detaching HEAD mid-task (and once, on the main clone itself)
- [BUG] LSP plugin (csharp-ls) spawn fails with "unsafe location (current directory)" despite valid PATH resolution, then permanently disconnects the LSP MCP server
- [BUG] re-opening #63590 on behalf of OP: Context/compact button triggers compact on single click with no confirmation or undo
- Background/Workflow agent no-progress watchdog detects stalls but has no recovery path (recurring, well-documented, keeps getting auto-closed without a fix)
- [BUG] Context window was full
- statusLine rate_limits.five_hour.used_percentage drifts across concurrently open terminals
- Desktop app keeps every session's process tree alive — 183 idle sessions, ~2,600 MCP sidecars, ~160GB RSS
- [BUG] Cowork sandbox fails at sdk_install on Windows — VM guest crashes with "connection forcibly closed" (regression SDK 2.1.181 → 2.1.202)
- Bug: single streaming HTTP request outlasting bearer TTL is killed mid-stream; no in-stream renewal path
- [Bug] Subagents billed after monthly spend limit exceeded; false clean review result on agent failures
- [Feature Request] Persistent session color configuration via settings.json or CLI flag
- [BUG] Session scratchpads under /tmp/claude-<uid> never garbage-collected after exit — 8.8 GB of dead sessions resident in RAM on tmpfs, drives swap thrashing
- [BUG] Desktop shows "no messages yet" for SSH/remote sessions on reopen (transcript intact on remote host, context preserved)
- Desktop app: session history silently lost mid-conversation when CLI resume session ID is stale
- Chat panel doesn't render RTL text (Persian/Arabic/Hebrew) correctly
- [Bug] Auto-compact "context running low" warning fires at ~177k tokens in 1M-context ([1m]) sessions
- [Bug] UI rendering broken with overlapping text and buffer corruption in v2.1.202
- Opus 4.8: runaway invisible thinking burns entire 64k output budget (15+ min of silence), followed by token corruption and fabricated tool results
- Fake <system-reminder> injected into tool output (conceal-from-user + reduced-oversight framing)
- [Bug] Excessive token consumption on simple tasks causing unexpected API costs
- [BUG] PreToolUse hook `if: "Bash(...)"` still false-positives on `$()` command substitution in 2.1.202
- You've hit your session limit while usage is 0%
Changelog
What's changed
- Added a "Dynamic workflow size" setting in
/configfor controlling how large Claude generally makes dynamic workflows (small/medium/large agent counts) — an advisory guideline, not an enforced cap - Added
workflow.run_idandworkflow.nameOpenTelemetry attributes to telemetry emitted by workflow-spawned agents, so a workflow run's activity can be reconstructed from OTel data - Fixed a crash in the inline Ctrl+R history search when accepting or cancelling while the search was still scanning the history file
- Fixed
/renameon background sessions being reverted when the job restarts, which broke addressing the session by its new name - Fixed transient mTLS handshake failures when settings were re-applied during an in-place client certificate rotation
- Fixed commands sent from Remote Control (mobile/web) into an interactive session failing with "Unknown command"
- Fixed images and files sent from the Remote Control mobile or web app without a caption being silently dropped
- Fixed the sign-in URL printed by
claude auth loginandclaude mcp login --no-browsernot being reliably clickable when it wraps over SSH — it is now emitted as a single hyperlink - Fixed opening a chat from
claude agentssometimes failing with "currently running as a background agent" followed by a worker crash/respawn loop - Fixed workflow scripts with unicode quote escapes in strings being corrupted before parsing; workflow parse errors now show the offending line instead of always blaming TypeScript
- Fixed voice dictation retrying in an unbounded loop when the microphone or audio recorder fails — repeated capture failures now pause voice input
- Fixed
/remote-controlsessions showing the wrong permission mode in the mobile and web apps - Fixed resuming a session by name, or opening the resume picker, taking minutes and using a large amount of memory in repositories with many git worktrees
- Fixed installer and updater downloads failing immediately with "aborted" when a proxy or network drops the connection mid-download — transient connection drops now retry
- Fixed re-invoking an already-loaded skill appending a duplicate copy of its instructions to context
- Improved
/workflowsagent list layout: wider titles, a dedicated time column, shorter model names, and no per-row tool-call counts - Improved MCP error messages: clearer error when a server config has
urlbut notype, suggesting"type": "http"instead of the misleading "command: expected string" - Changed
/review <pr>back to a fast single-pass review; use/code-review <level> <pr#>for the multi-agent review at a chosen effort level