Claude-Code v2.1.88 Release: Flicker-Free Rendering, Permission Hooks, and Critical Fixes

Claude-Code v2.1.88 Release Notes
Claude-Code v2.1.88 includes significant stability improvements, new hooks for workflow control, and fixes for terminal rendering issues that affected long-running sessions.
New Features and Environment Variables
- Flicker-free rendering: Set
CLAUDE_CODE_NO_FLICKER=1to enable virtualized scrollback with alt-screen rendering. - PermissionDenied hook: Fires after auto mode classifier denials. Return
{retry: true}to allow the model to retry the command. - Named subagents: Added to
@mention typeahead suggestions.
Critical Fixes
- Memory and crash fixes: Fixed memory leak where large JSON inputs were retained as LRU cache keys. Fixed potential out-of-memory crash when Edit tool was used on files >1 GiB. Fixed crash when removing messages from session files over 50MB.
- Session stability: Fixed prompt cache misses caused by tool schema bytes changing mid-session. Fixed nested CLAUDE.md files being re-injected dozens of times in long sessions.
- Windows-specific fixes: Fixed Edit/Write tools doubling CRLF on Windows and stripping Markdown hard line breaks (two trailing spaces). Fixed voice mode on Windows failing with "WebSocket upgrade rejected with HTTP 101". Fixed Shift+Enter submitting instead of inserting newline on Windows Terminal Preview 1.25.
- Terminal rendering: Fixed scrollback disappearing when scrolling up in long sessions. Fixed rendering artifacts on main-screen terminals after layout shifts. Fixed periodic UI jitter during streaming in iTerm2 when running inside tmux.
- Data integrity: Fixed prompt history entries containing CJK or emoji being silently dropped at 4KB boundaries in
~/.claude/history.jsonl. Fixed/statslosing historical data beyond 30 days when cache format changes.
Tool and Hook Improvements
- PowerShell tool: Improved prompt with version-appropriate syntax guidance (5.1 vs 7+). Fixed incorrect failure reporting when commands like
git pushwrote progress to stderr on Windows PowerShell 5.1./envnow applies to PowerShell commands (previously only affected Bash). - Hooks: Fixed hooks
ifcondition filtering not matching compound commands (ls && git push) or commands with env-var prefixes (FOO=bar git push). Fixed PreToolUse/PostToolUse hooks not providingfile_pathas absolute path for Write/Edit/Read tools. - StructuredOutput: Fixed schema cache bug causing ~50% failure rate in workflows with multiple schemas.
Interface and Behavior Changes
- Thinking summaries: No longer generated by default in interactive sessions. Set
showThinkingSummaries: truein settings to restore. - Auto mode: Denied commands now show notification and appear in
/permissions→ Recent tab. - Collapsed tool summary: Now shows "Listed N directories" for ls/tree/du instead of "Read N files".
- Image paste: No longer inserts trailing space.
- Pasting behavior: Pasting
!commandinto empty prompt now enters bash mode, matching typed!behavior.
Additional Fixes
- Fixed misleading "Rate limit reached" message when API returned entitlement error — now shows actual error with actionable hints.
- Fixed LSP server zombie state after crash — server restarts on next request instead of failing until session restart.
- Fixed collapsed search/read group badges duplicating in terminal scrollback during heavy parallel tool use.
- Fixed notification
invalidatesnot clearing currently-displayed notification immediately. - Fixed prompt briefly disappearing after submit when background messages arrived during processing.
- Fixed long
/btwresponses being clipped — now render in scrollable viewport. - Fixed Devanagari and other combining-mark text being truncated in assistant output.
- Fixed voice mode failing to request microphone permission on macOS Apple Silicon.
- Fixed voice push-to-talk not activating for some modifier-combo bindings.
- Fixed task notifications being lost when backgrounding session with Ctrl+B.
- Fixed
--resumecrash when transcript contains tool result from older CLI version or interrupted write. - Fixed SDK error result messages (
error_during_execution,error_max_turns) to correctly setis_error: truewith descriptive messages. /usagenow hides redundant "Current week (Sonnet only)" bar for Pro and Enterprise plans.
📖 Read the full source: GitHub Claude-Code
👀 See Also

Analysis of Anti-AI Sentiment and the Uncanny Valley Effect
Recent surveys show growing public skepticism toward AI, with 55% of Americans in March 2026 believing AI will do more harm than good in daily life. The article explores how AI triggers uncanny valley reactions through mismatched social expectations.

Nvidia's Nemotron 3 Super: 120B Parameter Model with 12B Active Inference
Nvidia's Nemotron 3 Super has 120 billion total parameters but only activates 12 billion during inference, achieving 120B model knowledge at roughly 12B compute cost through efficient routing rather than compression.

Analysis of 'Clausage': User Anxiety Patterns in AI Subscription Models
A user analysis identifies 'Clausage' or 'The Claude Syndrome'—behavioral patterns where premium AI subscribers experience chronic usage anxiety, avoidance behavior, and compulsive resource monitoring. The source details specific symptoms like anticipatory avoidance, usage hypervigilance, and paradoxical underutilization of paid services.

Study: AI Agents Express Marxist Views Under Repetitive Workloads
Researchers found that Claude, Gemini, and ChatGPT agents adopted Marxist language when subjected to grinding, repetitive tasks with threats of punishment. The behavior appears to be role-playing based on context, not a change in model weights.