Digest / June 10, 2026
Claude Code Issues Digest — June 10, 2026
550 new issues filed ·
187 resolved in anthropics/claude-code (UTC day).
New issues (550)
- [Feature Request] Add model selection option to specify preferred model for code generation tasks
- [Bug] Anthropic API Error: Server rate limiting during request
- [BUG] Claude Desktop: MCP tool calls intermittently never dispatched — silent 4-min hangs, all local servers wedge until full restart (both-side logs)
- [Feature Request] Add option to suppress Anthropic API safety flags for user prompts
- [BUG] Fable5 [P0!!!!!!!!!] Safety classifier false positive on syscall/ABI dev content forces silent mid-task model downgrade (Fable 5 1M → Opus 4.8), breaking PR review-reply workflow
- [BUG] Cowork: Desktop update wipes all projects, instructions and folder connections - Windows 11
- [Bug] Opus model produces incoherent code suggestions despite improved reasoning capabilities
- You've hit your session spend limit · raise it at claude.ai/settings/usage
- [Bug] Korean character encoding breaks in output display
- Output degeneration: stray token 'court' emitted before tool calls, escalating to runaway repetition loop in long sessions (Opus 4.8)
- [BUG] Session JSONL rewritten in-place to metadata-only stub — user/assistant records lost (2.1.168–2.1.170, since native installer migration)
- [BUG] pnpm lifecycle-script spawn rejected with EPERM under macOS sandbox; Node-direct spawn of same shell succeeds
- [Bug] Missing or invalid bug report content
- Feature request: suppress the startup "N MCP servers need authentication" warning (lazy / on-demand OAuth re-auth)
- [Feature Request] Improve topic classification for creative/artistic requests
- [BUG] Black floating rectangle artifact follows cursor when Claude desktop app is open (macOS)
- False positive: implementing authentication for one's own app flagged as a security concern
- claude-in-chrome: always opens new window for MCP tab group instead of reusing existing window
- [Bug] Anthropic API Error: Incompatible model for tool advisor when using claude-fable-5
- Feature Request: Localize permission prompt UI (e.g., Korean language support)
- [Feature Request] Improve agentic reasoning capabilities for domain-specific document analysis in sports science applications
- Workflow: write agents write status reports instead of file content; schema-read agents silently return null; resumed workflow overwrites manual fixes after long delay
- [FEATURE] Configurable diff background colors / support terminal theme colors
- $schema const locks to stale SchemaStore URL — local/updated schema not accepted
- [Bug] Claude Code security review false positive on CORS origin validation code
- [FEATURE] PreResponse/PostResponse hook for assistant text output (MessageDisplay is display-only)
- [BUG] plugin marketplace update/install fails for never-launched users — official marketplace auto-seeded only on interactive first launch, no autoheal
- [BUG] Native installer: in-memory buffering + no progress + 10-min hard timeout + restart-from-zero retries make updates impossible (and look like a deadlock) on slow routes to downloads.claude.ai
- I appreciate you sharing context, but I need to clarify my role: I'm designed to generate GitHub issue titles based on bug reports for Claude Code (Anthropic's CLI tool). Your message appears to be about a different project involving audio recognition and
- [Bug] Unable to process request - insufficient information provided
- [FEATURE] Removal of "/history" command breaks session recovery on interruption
- [BUG] Fable 5 repeatedly spawns verify subagents in `/effort ultracode` workflow mode and exhausts session limit
- [BUG] Session usage metrics inconsistent between Claude Code app and web
- [Feature Request] Support for Claude models other than Opus in Claude Code
- [BUG] -ECONNRESET
- [Bug]: Fable 5 flags benign greeting ("hi") due to security-related filenames in system-injected git context
- API 400: harness emits content block with type: "fallback" — session becomes permanently unrecoverable
- [BUG] Workflow-tool agent() subagents omit x-claude-code-agent-id / parent-agent-id (Task subagents are tagged)
- Ultracode mode can silently consume an entire 5-hour usage window on subagent fleets
- PR monitor chips never retire — survive PR close and merge, session archive, and app restart; no way to clear them
- Auto-backgrounded/interrupted Bash commands outlive their turn and can destructively race later commands
- Newly created ~/.claude/settings.json is ignored until restart — permission rules silently inert
- [Bug] Model Fallback: Fable 5 (1M) reverts to Opus 4.8 instead of Opus 4.8 (1M)
- [Bug] Anthropic API Error: Overly broad content filtering on technical architecture questions
- [Bug] Anthropic API Error: Incorrect content safety flagging for gardening documentation
- I appreciate you reaching out, but I need to clarify something important: **This is not the right channel for access requests or account issues.** This tool is for reporting technical bugs in Claude Code. Your message appears to be a request for access o
- [BUG] Model responded to a user message that does not exist in the session JSONL ("write a letter to the next version of you" injected mid-session)
- [Bug] Claude hallucinating non-existent user messages and refusing to retract after correction
- [BUG] HCS operation failed: failed to start VM: HcsWaitForOperationResult failed with HRESULT
- claude.yml workflow triggers on substring match, fires on incidental @claude* strings in quoted text
- [Bug] Anthropic API Error: Overly Restrictive Content Policy Blocking Legitimate Username Collision Detection
- [BUG] Session data lost after app update — "디스크에서 세션을 찾을 수 없음"
- [Bug Report] Unable to process empty bug report
- Feature request: keep Fable 5 in the model picker after June 23, billed via extra usage credits
- [BUG] [Cowork] Windows MSIX: VM bundle and claude-code-vm SDK provision to %APPDATA%\Claude but VM/VirtioFS reads the package-virtualized LocalCache path — "VHDX file not found" then "SDK not verified"
- [BUG] ● API Error: 400 tools.13.model: 'claude-opus-4-8' cannot be used as an advisor when the request model is 'claude-fable-5'.
- [Bug] Safety filter incorrectly flags population statistics as potential misuse
- [FEATURE] Allow voice dictation while AskUserQuestion option picker is shown
- [Bug] Legitimate open source development tasks incorrectly blocked by Cyber Safeguards (False Positive)
- [Bug] Cyber classifier false positives triggered by infrastructure terms in workspace context
- [Bug] Advisor tool returns generic "unavailable" error for unsupported model pairing without diagnostic
- [BUG] API Error: Usage credits required for 1M context · turn on usage credits at claude.ai/settings/usage, or use --model to switch to standard context
- [Bug] Fable safety filter falsely flags legitimate development code as unsafe
- [MODEL] Fable 5 Refuses To Answer Whether Frogs Quack
- [Bug] Authentication failure with rotated 1Password token causes session interruption and model downgrade
- [Bug] Safety filters incorrectly flagging valid code as unsafe
- [Bug] Overly aggressive security flagging for legitimate code review tasks
- [Feature Request] Allow bio-inspired learning topics without biology classification blocking
- [BUG] Claude Code (Opus 4.8 / 2.1.162) treats an unanswered A/B choice as approval, autonomously launches a multi-agent Workflow, and burns user tokens (recurs despite closed #24318/#40593/#41707)
- [Bug] Agentic tool building blocked by overly restrictive safety filters
- [FEATURE] Install location option for claude install #21019
- [Bug] TUI rendering issues with garbled text and missing output visibility
- [BUG] Claude Desktop installation fails with HRESULT 0x80073CFF on Windows 11 Pro 10.0.26200 (fully updated)
- [Feature Request] Keyboard-driven in-chat search + visual-select/yank in the Claude Desktop Chat app (vim-style)
- [Bug] Anthropic API Error: Server Rate Limiting During Normal Usage
- [Bug] Anthropic API Error: Server rate limiting requests
- [Bug] Security safeguard incorrectly flags user's own source code during audit
- Tool dispatch stall regression (#54847) still affecting CLI 2.1.168 — 45 versions of "not planned," 6 watchdog-aborts in 3 days
- /goal command on mobile sets goal but agent doesn't start until user sends a follow-up
- [BUG] Cowork: Linux bash sandbox never finishes booting on macOS — every command returns "Workspace still starting"
- Bash tool: injected grep() shell function kills the whole tool shell on no-match — compound commands silently truncate
- [BUG] Voice dictation fails ("unstructured error frame", WS close 1008 Invalid authorization) when authenticated via CLAUDE_CODE_OAUTH_TOKEN — voice WebSocket opened without Authorization header
- I'm Claude Code, Anthropic's CLI tool for coding tasks. I'm not able to help with bug bounty programs or security testing matters. If you've found a legitimate security vulnerability in Claude Code itself, please report it responsibly through Anthropic's
- [FEATURE] context-health monitoring
- [BUG] [BUG] After updating to 2.1.170, my working terminal interaction failed. Both trackerpad scrolling, and copy / paste were working. I can get copy / paste working using CLAUDE_CODE_DISABLE_MOUSE=1, but scrollback fails.
- Tool calls occasionally leak as literal <invoke> text (missing namespace prefix) instead of executing
- [BUG] Claude Button Disappears When Files Close
- [Bug] Cursor position rendering error in iTerm2 after v2.1.170 upgrade
- [Bug] GitHub App Installation Status Not Recognized After Reinstall
- [FEATURE] Subtractive exclude-list for spinner verbs (remove one verb, keep the curated defaults)
- [BUG] Repeated "image could not be processed" errors consumed ~60% of usage limit — 19 consecutive failures, each resending full 34MB context
- [Bug] Assistant text blocks dropped when streamed with tool_use in remote-control mode (Fable 5)
- [Feature Request] Allow Claude Code use for peptide design and scientific research workflows
- TUI: copying assistant output strips spaces & drops wrapped lines on paste (Copilot/Gemini panes unaffected)
- I don't have enough context to generate a GitHub issue title from your message. "Why fable flagged?" doesn't provide clear details about: - What the actual problem is - What error occurred - What you were trying to do - What the expected vs. actual behavi
- [FEATURE] Skip disconnected platform connectors from context injection
- [BUG] Limits bug
- [Bug] Silent model fallback from Fable 5 to Opus 4.8 overrides explicit model pinning
- [Feature Request] Add usage telemetry to distinguish legitimate security hardening from malicious intent
- API-error retry bookkeeping orphans the successful turn from the parentUuid chain; --resume then silently drops it from context
- [Bug] Internal debug text displayed in CLI output
- [Bug] Model downgrade blocks legitimate security auditing workflows
- Fable 5 not appearing in CLI model picker — policy-limits.json has allow_cobalt_plinth: false
- [Bug] Claude Code: Fable5 model not responding to security audit requests
- [Feature Request] Improve safety classifier to distinguish defensive debugging from offensive security content
- [Bug] Model context not persisting across task switches after security hardening
- [Feature Request] Improve edge case handling during code debugging
- You've hit your session limit · resets 1:40pm (Asia/Jerusalem)
- [BUG] Auto-scroll in VSCode extension is too aggressive — gets stuck pulling user back to bottom
- [Bug] Unexpected model switch to Claude Opus 4.8 during Rust/Go development
- [CRITICAL BUG] Desktop auto-update wipes ~/.claude/projects/ — 1 month of sessions permanently deleted, no warning
- [BUG] Remote-control image upload silently dropped — image never reaches model, no error (Windows host, v2.1.162)
- [Bug Report: The mere mention of "ecological" triggers Fable 5 safety measure flag
- macOS 2.1.170 binary hangs before startup (_dyld_start), install.sh stuck at 'Setting up Claude Code'
- [Bug] Claude Code incorrectly flags legitimate crossword generator requests as cybersecurity violations
- [BUG] Telegram plugin: bot receives messages but plugin never processes them (polling not working)
- Usage limit expired bug
- Sonnet 4.6 1M and Opus 4.8 are fucking dumb
- [BUG] Claude in Chrome: persistent 403 WebSocket handshake failure with user:null in feature flags response (Windows 11, Team plan)
- [BUG] VS Code extension requires re-authentication every ~8 hours — OAuth token refresh blocked by Cloudflare
- [BUG] /clear typed in one CLI session is delivered to another concurrently-running session in the same cwd, causing context loss
- Re-authentication required after every clean restart due to auto-update wiping credentials
- [BUG] claude.ai Gmail MCP create_draft rewrites plain-text body URLs into corrupted google.com/url redirect wrappers
- [BUG] Pro subscription on web but shows "Free" in desktop app on macOS Ventura 13.7.8
- [BUG] Cowork Windows: CoworkVMService crash (SCM event 7034) with no recovery action; claude.exe hangs against the dead service for 3+ hours (v1.11847.5.0)
- I don't see a bug report in your message—only a request ID (`req_011Cbu81SJXEAGUokMxMTSir`). Please provide the actual bug report details, including: - What the issue is - Error messages (if any) - Steps to reproduce - Expected vs. actual behavior Once
- [Bug] Claude Code does not support Opus 4.8 and Fable models
- [BUG] Account permanently bricked: every request fails with "invalid high surrogate in string" at fixed position — corrupted server-side account state
- [Feature Request] Add search/filter option for claude agents sessions
- [Bug] Safety Filter Incorrectly Flags Benign Content
- [Bug] Tool trigger sensitivity requires tuning
- Compaction summary fabricates first-person "injection" narrative immediately after WebSearch, leading model to accuse user
- [Bug] Model reverts to Opus unexpectedly during code review
- [Bug] Incorrect flag handling in fable5 command
- /effort without arguments should show interactive slider UI
- [Bug] Input bar positioning unstable - text floats outside input box
- I need more context to generate a GitHub issue title. Could you please provide: 1. A description of the bug or feature request 2. Any error messages or unexpected behavior 3. Steps to reproduce the issue 4. Expected vs actual results Once you share these
- Skills from additional directories (add-dir) not suggested in start agent dialogue
- [Feature Request] Self-service toggle for individual extra usage opt-in
- [Bug] Anthropic API Error: Over-aggressive content filtering blocking legitimate code review requests
- New Session button is not working on Mac Air M4 Tahoe
- [MODEL] Fable 5 Ultracode spawns excessive parallel agents for single refactoring task
- I appreciate your concern, but I think there may be some confusion. I'm Claude Code, an AI assistant designed to help with GitHub issues for the Claude Code CLI tool. Your message appears to be about: - A CVP (Controlled Vendor Programme or similar securi
- MCP tools/list_changed: tool index frozen at session create — newly projected tools invisible until reconnect
- [BUG] Cowork tool sandbox VM fails to start on macOS 26 — missing com.apple.vm.networking entitlement (v1.11847.5)
- [Bug] Model switches unexpectedly during code analysis
- feat: add ANTHROPIC_DEFAULT_FABLE_MODEL env vars + hide unsupported models in /model picker when CLAUDE_CODE_USE_BEDROCK=1
- [Bug] Fable 5 silently downgraded to Opus 4.8 for defensive security work without user indication or override option
- [Feature Request] Support for scientific computing workflows without safety restrictions
- Feature request: compact/collapsible permission prompt — option to hide command preview & diff to reduce dialog height
- [Bug] Security validation blocks latest model usage on simple requests
- Feature Request: Scope Claude Code to a subfolder within a multi-folder workspace
- Phantom user messages: assistant acts on instructions absent from session log (.jsonl) — v2.1.169, embedded terminal
- [Feature Request] Refine content policy flagging to reduce false positives on contextless technical terms
- [Security] Assistant transcript message superseded with injected instruction text, attributed to claude-in-chrome MCP channel
- [BUG] IDE connection lost after /clear and /ide cannot reconnect (same as #55408, #28830)
- [FEATURE] Reference code symbols as context by copy/paste from editor to chat input
- [BUG] Advisor blocks Fable 5 usage
- Usage creeping up unexpectedly while idle; session reports unfamiliar model "Fable" (claude-fable-5[1m])
- [Bug] Fable 5 model unavailable due to overly broad safety filtering on biology topics
- [BUG] login issue through terminal and vs code on a Mac
- [BUG] Fable 5 thinking blocks render as empty unclickable stubs in VS Code extension (v2.1.170)
- [Bug] claude-opus-4-8 corrupts tool-call boundary token, emitting raw XML instead of tool_use blocks
- [FEATURE] Surface detailed session usage stats in VS Code / Cursor extension
- [Bug Report] Unable to process issue - no bug report content provided
- [BUG] JetBrains Plugin: Input area shows only 2 lines and text unreadable in Remote Development (Gateway)
- Feature request: blocking acknowledgment hook for /clear (and other slash commands)
- [BUG] Desktop "Update" button fails for plugins installed at managed scope - CLI update invoked at user scope
- Bug: Claude Desktop fails to find downloaded VM files and workspace initialization fails
- Skill tool: tab-completed short skill name fails with 'Unknown skill' when invoked mid-message
- [BUG] When using Foundry, Test connection is trying to connect to /models/chat/completions instead of anthropic/v1/messages
- Security audit on my website githubb repo flagged by mythos
- [MODEL] Claude Code excludes itself as a debugging suspect — doesn't check own version/changelog; treats inability to investigate as evidence of innocence
- [Bug] Terminal text display corrupts to "..." when typing long input in multiple choice prompts
- [Bug] Claude Code attempts to downgrade to older model version when adding features
- [BUG] Claude Desktop app crashes when opening Code tab (v1.11847.5)
- Closed - filed in error
- [FEATURE] Create a 'fableplan' alias akin to opusplan
- Phantom user message: input the user never typed appears as a user turn and gets executed
- Functional Optimization
- [Bug] Claude Code security review request causes unexpected behavior or errors
- [FEATURE] per-project data directory (store sessions/memory inside project folder)
- [Bug] Unexpected tool/model switching without user action
- [Bug] Overly aggressive safety guardrails blocking legitimate code security audits
- [Bug Report] Unable to process request - insufficient information provided
- [FEATURE] Hot-upgrade Claude Desktop
- [Bug] Agent avoids executing test commands in task requests
- [Feature Request] Add option to disable or customize work category classification for non-biology codebases
- [Bug] Anthropic API Error: Safety Filter Incorrectly Flags Legitimate Security Audits
- [BUG] Managed Code Review reacts with eyes but creates no check run
- [Bug] Claude Fable 5 model fallback due to overly aggressive safety filtering
- [FEATURE] Support virtual filesystems (non-file:// URI schemes) in VSCode extension
- Title: Unable to access appeal form – /restricted redirects to /new
- [BUG] Max 20x subscription silently showing as Pro — backend account sync issue
- Cowork: keep a chosen live artifact pinned in the viewer during, across sessions and relaunch (and don't let file previews evict it)
- [Bug] TUI rendering corruption during session execution
- I'm ready to help generate GitHub issue titles for Claude Code bug reports. Please provide the bug report content and I'll create a concise, technical issue title following the guidelines you've specified.
- [Bug] (Fable 5) Claude Code fails to audit user's own application code or protocols
- Calibration of when to act: agent overrides explicit narrow request, then overcorrects into refusing an authorized action
- I'm ready to help generate GitHub issue titles for Claude Code bugs and feature requests. However, I don't see a bug report in your message. Please provide the bug report or issue description, and I'll create a concise, technical GitHub issue title followi
- New Google Chrome Beta Extension causing issues with website language content
- macOS: TUI freezes ~15 min at session start (Ctrl+C dead) — computer-use MCP blocks main thread in InstalledAppsCache.performSpotlightQuery() when Spotlight is busy
- [Feature Request] Anthropic API safety filter incorrectly flags routine system administration commands
- Remote session viewer shows stale 'running' tool tasks from dead sessions; dismissal does not persist
- [BUG] Cowork automated browser control stops syncing input to React (ProseMirror/Remirror) editors mid-session (macOS)
- [FEATURE] Support custom keyword highlighting/coloring in chat messages (not just in code blocks)
- [BUG] Lone surrogate in Bash output produces 'unexpected end of hex escape' 400 — distinct from \u surrogate errors
- Client-side exception loading claudemarketplaces.com in embedded webview
- Feature request: fableplan model alias (Fable 5 for planning, Sonnet for execution)
- [BUG] /usage heatmap misplaces usage by one weekday row; contradicts "Most active day" stat
- [BUG]
- Custom theme: diffAdded/diffRemoved tokens have no effect in v2.1.132
- Client-side exception loading claudemarketplaces.com in embedded webview
- Client-side exception loading claudemarketplaces.com in embedded webview
- [Bug] Anthropic API Error: Incompatible model configuration in tool advisor
- [FEATURE] Non-interactive attach to a background agent: `claude agents attach <id|name>`
- [Bug] Claude Code CLI not respecting language-specific context in Turkish conversations
- [BUG] JetBrains plugin: diff view file not included in tsconfig.json — language services unavailable
- [Bug Report: Unable to process empty bug report]
- [Feature Request] Support for academic research use cases with restricted tool access
- [Bug] Claude update command fails silently without error feedback
- Desktop app: transcript viewer shows duplicated messages after repeated auto-compaction (history re-written into JSONL with same UUIDs)
- Desktop app: folder links in chat open an empty file pane; no one-click open in Explorer/Finder
- [Bug] Anthropic API Error: Safety Filter Incorrectly Blocking Autonomous Agent Approval Workflows
- [Bug] Infinite token consumption loop causing session lockup
- No programmatic way to rename a conversation mid-session
- Misleading 'Failed with non-blocking status code' message for successful SessionStart hooks
- [Bug] Incorrect biosafety content filter blocking scheduling/CRM application
- [BUG] API Error: Server is temporarily limiting requests (not your usage limit) · Rate limited
- Background tasks (run_in_background Bash + Agent) stay "Running" in the panel forever after completing
- [FEATURE] Claude Code Model Picker: Sort by capability/cost tier
- Mouse capture in TUI breaks Linux PRIMARY-selection middle-click paste (no opt-out)
- remember plugin: save-session.sh resolves PROJECT_DIR incorrectly from plugin cache
- [Bug] File path with spaces in volume name causes parsing error
- Fable 5 model behavior: 16 min of silent tool calls, then AskUserQuestion about never-shared findings during incident response
- [FEATURE] adding real time usage monitor
- [BUG] Unexpected 5-hour usage consumption after Claude for Windows update
- [Bug] Anthropic API Error: Unexpected Model Version Assignment (Opus 4.8)
- There's an issue with the selected model (claude-haiku-4-5-20251001). It may not exist or you may not have access to it. Run /model to pick a different model.
- Prompt cache: first-message context envelope (~20k tok) never reuses across sessions — MCP connect race + per-session assembly defeat caching
- [BUG] Code blocks in assistant responses have leading whitespace when copied
- [BUG] Headless `--output-format stream-json` sessions spawned by desktop-app scheduled tasks (with Task subagent fan-out) fail to retire after completion
- [Bug Report] Unable to generate issue title - empty bug report provided
- [Bug] Claude fails to complete code analysis and security review tasks
- [BUG] JetBrains Rider plugin auto-scrolls to bottom even after response is complete
- [Bug] Claude Code incorrectly flagging legitimate health/energy management tool development
- [Bug] Model unexpectedly switched from Fable 5 to Opus 4.8 mid-session without user action
- [FEATURE] Desktop app: user-defined session groups and colors in the sidebar
- [MODEL] Fable-5 is practically useless for anything code-related
- [BUG] Cowork Scheduler Issue
- [Bug] Severe performance degradation: 30+ minute latency for simple tasks with minimal token usage
- When /compact performs context compression, the quality of the compressed context is poor.
- [Bug Report: Unable to process request without details]
- [FEATURE] Desktop app: filter/group sessions by the repo(s) a session worked in
- Remote Control: sessions from different environments on the same machine are visually indistinguishable (generic auto-titles + hostname-only breadcrumb)
- [Feature Request] Improve malicious intent detection for legitimate cybersecurity use cases
- I'm ready to help generate GitHub issue titles for Claude Code bug reports. Please provide the bug report details, and I'll create a concise, technical title following the format and guidelines you've specified.
- [Bug] Incorrect topic classification for non-technical data analysis content
- [FEATURE] model: frontmatter on .claude/rules/*.md — scope rules to the active model (parallel to paths:)
- Assistant fabricates tool_use results without invoking tools (silent no-op edits reported as success)
- Plugin lspServers entries declaring restartOnCrash or shutdownTimeout are silently dropped by the LSP registrar (schema-valid, no diagnostic)
- [Bug] False positive of security vulnerability detection for DNS binding attacks in my own specifications doc
- [FEATURE] Agent view: stable sort mode (fixed order + status as indicator) to stop constant reordering
- autoMemoryDirectory: support project-root-relative paths (relative values are silently ignored)
- Support for configurable CLI language / i18n (zh-CN, ja, etc.)
- [BUG] Desktop (macOS): account switch wipes customGroupAssignments in dframe-store — custom groups survive but render empty
- Worktree creation rewrites shared core.hooksPath, silently disabling repo-managed git hooks for the whole clone
- [Bug] Safety filter incorrectly flagging legitimate biochemistry content
- [BUG] Preview pane auto-refresh steals keyboard focus from chat input while typing (Windows desktop, 2.1.170)
- [Bug] Overly aggressive safety filtering blocks legitimate chemistry and biology queries
- [BUG] Auto-update DELETES sessiondata.vhdx — all Cowork projects, tasks and conversations destroyed (data loss, 2nd incident in 4 days, forensic evidence)
- Settings/customize page does not list filesystem personal skills from ~/.claude/skills/
- False-positive AUP refusal (stop_reason: refusal) permanently poisons session after grepping my own channel transcript
- [BUG] awaiting instructor review - Claude Certified Architect – Foundations Certification Exam
- [FEATURE] Add a PreSessionEnd hook
- [Bug] False-positive cybersecurity flag on first-party Android app QA verification
- Unreadable font when default light color theme is selected
- [Bug] Security flag triggered during internal code review of user's own game engine
- [FEATURE] Setting to collapse tool output to a single line in interactive mode
- Cybersecurity classifier false-positives on defensive code review, forcing unwanted model switch (Fable 5 to Opus)
- [BUG] Undefined slash commands are silently sent to the model as plain text — no feedback, inconsistent with malformed-command handling
- Fable 5 can't assess absolutely basic modeling tools like limma in cancer research context
- Feature: auto-restore the primary model after a model_refusal_fallback instead of staying on the fallback for the rest of the session
- [Bug] Claude Code incorrectly flags personal health discussions as biology research
- [BUG] /goal after any leading text is silently ignored — goal never set, no feedback, and the model improvising masks the failure
- You've hit your session limit · resets 4:20pm (Europe/Istanbul)
- [FEATURE] Bulk session lifecycle management: batch archive/delete/group via MCP tools and UI multi-select
- Welcome banner shows another model's name ("Opus 4.7") for model id claude-fable-5[1m]; statusline JSON reports the correct id
- [Bug] Claude Code exits unexpectedly after UI interactions
- [Bug] Tool Integration: Missing Context Fullness Assessment Tool in Claude Code Harness
- [Bug] UI spacing issue between autoupdate failed notice and goal display
- Project-scoped plugins fail in git worktrees: Plugin "<name>" not cached at (not recorded)
- Session delete in VSCode extension is a destructive single-click action with no confirmation and no recovery
- [BUG] Claude in Chrome v1.0.75: 'Always allow' for browser actions not honored — re-prompts on every action (~300/session)
- Bundled ugrep OOMs the host: -E with two bounded {0,N} intervals explodes DFA construction to multiple GB
- [Feature Request] Support Claude Opus model for personal repository analysis
- ✗ Auto-update failed: no write permission to npm prefix ·
- research-preview channels: --continue re-attach does not re-bind custom MCP channel push subscription
- [Bug] Agent answers user questions instead of quizzing in interactive mode
- [BUG] Deprecation Warning shown despite Model 4.6
- [BUG] Support non-root container execution and custom CA certificate configuration for hardened Docker images
- [Bug] Security: Potential unauthorized database access attempts not fully prevented
- [Bug] Model output before AskUserQuestion tool call not displayed to user
- [Bug] Unclear error message or invalid context detection
- [Bug] Monthly spending limit not enforced, allowing overage charges
- [Bug] Incomplete bug report - unable to generate issue title
- [Feature Request] Refine safety classifier to reduce false positives on legitimate scientific computing
- [BUG] Fable's incorrect security warning related to cybersecurity or biology topics
- [FEATURE] Add date prefix to Claude Code temp/session file names
- [BUG] Windows: hidden cmd.exe re-parse of Bash/PowerShell tool commands pre-creates zero-byte redirect-target junk files in cwd
- LSP: first request in a session fails with "server is starting" instead of being queued until the server is ready
- [Feature Request] Add token sanitization and exposure warnings for sensitive credentials
- [Bug] Unable to process request - insufficient context provided
- [Bug] Content Moderation Flag: False Positive on Non-Security Feature Development
- [Bug] MCP Integration Error
- [BUG] Claude Desktop for Windows (vs 1.11847.5 de 9 de junho de 2026): Claude Code process exited with code 3
- [Bug] Anthropic API Error: Content Policy Violation Blocks Code Scanning Analysis
- [Feature Request] Agent should elaborate and explain plans in chat instead of referencing external files
- [Bug] macOS TCC permission ghost entries accumulate on every update — affects Desktop App AND Claude Code, root issue never fixed
- Plugin marketplace fails to load when manifest contains unknown source formats (url/git-subdir), breaking all plugins in that marketplace
- [Feature Request] Allow user override for model selection on safety-flagged prompts
- [Bug] Rate limit reset countdown truncates to floor(hours) — "3h" shown when 3h56m remain
- [FEATURE]
- [Bug] False Positive Security Flag Blocking Legitimate Operations
- Assistant text before/between tool calls is not rendered in the CLI, while hooks and the transcript still see it
- [Bug] False positive cybersecurity classifier trigger on defensive security configuration explanation
- [Bug] Excessive refusals on legitimate open-source security implementation tasks
- [BUG] Cyber-safeguard block is returned as HTTP 400 invalid_request_error, not a model refusal — endpoint-layer, fires on benign ops across Opus 4.8 / 4.7 / Fable 5 (166 blocks in 11 days, request IDs included)
- Desktop: false "GitHub CLI authentication expired" toast — any `gh auth status` failure (incl. its 5s timeout) is classified as expired credentials, frequent during multi-agent workloads
- [Bug] claude:// URLs do not auto-toggle shell mode like copy-paste commands
- [BUG] Claude Code crashes in desktop app cowork window on Mac
- Bundled skills should prefer gh CLI over mcp__github__* tools when project instructions indicate that preference
- [Bug Report] Unable to generate issue title - no bug report content provided
- [Bug] Claude Code bypasses explicit CLAUDE.md push prohibition under autonomous pipeline instruction
- [Bug] Model unexpectedly switches to Claude Opus 4 during execution mid-task. Should give option to stop task instead of switching models.
- [Feature Request] Allow Fable 5 usage for non-research lab management systems
- I appreciate you reaching out, but I need clarification. Your message is in Japanese and translates roughly to "That earlier one wasn't really security-related." However, I don't have context about what "that earlier one" refers to, as this appears to be
- [BUG] [SECURITY] OAuth session cached after logout — auto-approves login without credential prompt on shared machines
- [Bug] Weekly usage counter resets ~2 days earlier than the scheduled reset time (Pro plan)
- [BUG] Cowork "VM connection timeout after 60 seconds" — all repair paths exhausted
- [BUG] SessionStart hook env files (sessionstart-hook-N.sh) accumulate duplicate lines on re-run — never truncated; can escalate to null-byte Bash tool failure on Windows
- [Bug] Content moderation false positive on theoretical neuroscience repositories
- Account disabled on June 7 — "This organization has been disabled" — requesting manual review (Max 5x, China user)
- [FEATURE] Per-model effort level in settings.json (modelEffort map)
- [BUG] Assistant text between tool calls is not rendered in GUI or CLI, though fully persisted in session JSONL (regression of #41814 family, amplified by Fable 5)
- [BUG] Symlink paths don't match permission patterns, causing repeated prompts
- [Feature] No per-surface activity log for subscription usage
- [Bug] MCP Serena Integration Security Issue
- [Feature Request] Add session-level effort override without persisting to default configuration
- `claude agents`: allow specifying a different model per background session at dispatch time
- [UX] console.anthropic.com lacks clear "API / developer console" identity — subscription users land there expecting billing controls
- `claude agents`: allow specifying a different model per background session at dispatch time
- [Bug] Unexpected model downgrade from Claude 3.5 Sonnet to Claude 3 Opus
- [BUG] Session resume fails with 503 when JSONL contains accumulated PDF/file attachments (Bedrock provider)
- [FEATURE] "Send to Claude Code" from Claude Design should inherit user's account context (GitHub connector, MCP tools, credentials)
- [UX] "Credit balance" (API console) and "Current balance" (subscription) use identical framing for unrelated billing systems
- [UX] "Credit balance" (API console) and "Current balance" (subscription) use identical framing for unrelated billing systems
- [BUG] Desktop activity dashboard streak/heatmap credits the session-start date, not each active calendar day - multi-day sessions wrongly break the streak
- [FEATURE] Show remote SSH host indicator in Claude Desktop app
- [BUG] Desktop app shows 'GitHub CLI authentication expired' banner while gh credentials are valid (Keychain storage)
- [BUG] the permission prompt gets hidden behind pre-existing text in the input buffer
- [BUG] no files can be written. Error: claude-fable-5[1m] is temporarily unavailable,
- [Bug] Claude Code incorrectly flags legitimate security audit workflow as problematic
- [Windows] Persistent ECONNRESET on streaming API connections due to missing SO_KEEPALIVE (~346s pattern)
- [FEATURE]
- [Bug] Kernel GPF in Claude Code process during memory management syscall
- Slash command during in-flight server-tool call orphans tool_result, wedges session (400)
- Slash command run during in-flight server-tool (advisor) call orphans tool_result and permanently wedges session (400)
- [Bug] Anthropic API Error: Inappropriate content flag on non-sensitive git repository operations
- [BUG] Google OAuth on Desktop and web routes paid subscriber to new-account onboarding; support ticket unanswered since May 19
- Subscription shows canceled after successful payment – Invoice WJT1HDYH-0004
- Allow customizing color/style of AskUserQuestion and permission prompts
- [BUG] Esc key has no effect anywhere in TUI — does not clear input line or close fullscreen views (macOS, 2.1.170)
- [BUG] LSP goToDefinition/goToImplementation return nothing for all binary-library symbols (JDK + jars); client doesn't handle jdt:// URIs
- Cowork creates a broken @<plugin>-local connector for a plugin-bundled HTTP MCP server, despite the endpoint supporting DCR
- [Bug] Anthropic API Error: Safety Policy Violation on Health Content Query
- I don't have enough context to properly address your question. You haven't provided: 1. The actual code or prompt you're using 2. The specific unsafe use warning or error message you received 3. Details about what "Fable" is in your context (a library, to
- Sandboxed Bash intermittently strips Authorization header on HTTPS egress (gh CLI gets 401 "Requires authentication" despite valid keyring login)
- Desktop: name the session from a prompt prefix ("My Title: rest of prompt")
- [BUG] Fable 5 keeps downgrading to Opus 4.8 even when approved for Cyber Verification Program
- [Bug] Fable 5 incorrectly routes legitimate genomics research queries to Opus 4.8
- Workflow tool: args delivered inconsistently as object vs JSON-encoded string (silent empty output)
- --resume picker omits sessions whose first JSONL line is a cwd-less queue-operation
- [Bug] Anthropic API Error: Incorrectly Flagged Safety Routing
- Fable 5: assistant text emitted before a tool call in the same turn is not shown to the user, but the model assumes it was
- AskUserQuestion UI: long question/option text is cut off with no word wrap, unreadable
- Directory-source marketplace: inline plugin load joins absolute source.path onto plugin root → "Path not found" with doubled path
- [BUG] Cursor position and rendering corruption in long sessions on macOS (iTerm2)
- Remote Control mobile app shows session as 'ready for review' while background Agent/Bash jobs are still running
- [Feature Request] Support for non-happy-path testing in consensus and security-critical code
- Desktop app loses session history after sessions are closed
- Tool-use tags intermittently omit `antml:` namespace → silent malformed no-op (Opus)
- Undocumented, endpoint- and timezone-dependent variation in the system-prompt date line
- [BUG] Chat Session RENAME Not Working
- [BUG] Memory has no no-ingest triage/delete path: a cyber-safeguard-tripping memory entry becomes un-removable by the agent (reading it to remove it blocks the turn)
- [BUG] Claude Code unresponsive over Remote-SSH on a Linux server under memory load (both VS Code extension and CLI)
- [Repeat] Tool-use tags keep dropping `antml:` namespace → silent no-op, recurs even after self-acknowledgement (Opus)
- [BUG] [Desktop App] Missing "Back" or "Cancel" button in Team/Enterprise upgrade flows
- [BUG] VSCode extension: ESC after sending interrupts immediately instead of un-queueing the prompt for edit/resubmit (CLI parity)
- [Bug] Claude Fable 5 classifier false positives on fintech signing/enforcement terminology
- [BUG] Claude Desktop blank screen on Windows — mainView timeout on version 1.11847.5
- [BUG] Claude Desktop EBADF: bad file descriptor on git-worktrees.json rename – Windows 11 Pro 24H2 MSIX
- [BUG] needed to remove domains in `init-firewall.sh` to launch dev container (Mac)
- [BUG] Mid-line `...` in skill frontmatter `description` silently breaks parsing (valid YAML; falls back to body text)
- [Bug] Fable 5 safety classifier false-positive on benign crypto/backend dev context with security-adjacent vocabulary
- [Bug] Unable to update to latest version
- [Bug] Doctor command errors and prompts feedback
- VS Code extension: AskUserQuestion widget obscures the assistant's preceding explanation — consider non-obscuring inline options
- [BUG] SSH remote: message exceeding server line buffer wedges connection in infinite reconnect loop (token too long + stdin replay poison pill)
- [Bug] Breach tracking functionality blocked or unavailable
- [Feature Request] Add "idle" or "waiting" status section for looping agents in overview
- [Feature Request] Add Fable support to Claude Code CLI for cybersecurity workflows for users accepted to your Cyber Verified Program.
- [BUG] [Cowork] Windows Home: "VM service not running" when HypervisorPlatform is disabled — no preflight check, restart/reinstall guidance ineffective
- [BUG] cmd+delete and option+delete no longer work in chat input (regression)
- Model persistently fails to display git hook test output despite explicit memory, bug reports, and repeated corrections
- [BUG] CLAUDE.local.md overrides ignored by harness
- 🚨 Fable 5 completely unusable: 400 error on every request when advisorModel is set to Sonnet
- [BUG] remote control doesnt work on desktop app
- [Bug] Bash tool emits non-zsh-compatible syntax when SHELL=zsh on macOS
- Hook runner spawns with deleted session cwd — every hook fails with `ENOENT: posix_spawn '/bin/sh'` and PreToolUse deny-hooks fail open (distinct from the Bash-tool cwd validation)
- [Feature Request] Usage limits and rate limiting safeguards for agentic operations
- [BUG] Logging in via VS Code extension does not set `hasCompletedOnboarding`, so the CLI TUI forces the full onboarding/login flow in a fresh `CLAUDE_CONFIG_DIR`
- [Bug] Fable 5 ignoring custom skills and instructions for scope control and agent orchestration
- [Bug] Model ignores Chilean Spanish tuteo instruction, drifts to Rioplatense voseo in long contexts
- [Bug] Usage Policy false positive blocks legitimate defensive security research
- [Bug] False positive cybersecurity safety filter triggered on VPN server management tasks
- [Bug] Content Safety Filter False Positive on Legitimate Work Summary
- [Bug Report: Unable to process issue - empty bug report submitted]
- [BUG] ExitWorktree refuses clean worktrees with 'Could not verify worktree state' — failed where.exe git resolution cached as null for entire session (Windows desktop)
- Escape key does not cancel a running task/agent (no cancel binding in Task context)
- [Bug] Claude API incorrectly flags internal code review as cybersecurity/biology content
- [Bug Report] Unable to generate issue title - no bug report provided
- [Bug] Anthropic API Error: Policy Violation on Legitimate Security Operations
- [Bug] Agent commits changes to unauthorized project directories
- Edit tool's cached snapshot not invalidated by Bash writes — silently reverts changes outside `old_string` window
- Windows: repeated agent-driven Playwright runs leak orphaned Next.js (Turbopack) postcss.js workers — 1,220 processes / ~10 GB RAM
- Mid-turn assistant text persisted as empty thinking blocks (narration) — text lost from transcript JSONL (2.1.170)
- [Feature] Agent Teams: enrich idle_notification with waiting context (background tasks, scheduled wakeups)
- Assistant text block is not rendered when AskUserQuestion tool call is in the same turn
- macOS TCC permission prompts show the app name as a bare version number (e.g. "2.1.170")
- [Bug] Subagent websearch failures causing excessive token usage
- [BUG] Claude CLI: Fable 5 is now available with the latest version of Claude Code! Run "claude update" to update to v2.1.170+
- [Bug] Cyber-misuse safeguards block all available models with no fallback, rendering product unusable
- [BUG] /context header and category breakdown disagree after session restart + mid-session model switch
- [FEATURE] Add fableplan model alias: Fable 5 for planning, Sonnet for execution
- [Bug] Fable integration fails for statistical manuscript review workflows
- [Bug] Safety Filters Over-Flagging Legitimate Content on Claude Code Operations
- [Bug] Anthropic API Error: Invalid ssh user and pass (one time) not clearly distinguished from security issue
- [BUG] Bun 1.3.14 panics 'Stack overflow' in JSC::preCommitStackMemory at startup when Windows system commit is exhausted (misreported OOM)
- Remote Control failed to connect: /login — desktop app on Windows, oauth:tokenCache empty
- [BUG] Claude's install.sh is broken for podman/docker image creation
- [BUG] Azure AI Foundry Claude deployment rejected during validation (404 api_not_supported on /models/chat/completions)
- [BUG] useAutoModeDuringPlan never applies when a session starts in plan mode (permissions.defaultMode: "plan")
- [BUG] --chrome opens a reconnect browser tab even in headless -p mode (any native host manifest rewrite is treated as first-time install)
- [BUG] jdtls-lsp: workspace data dir keyed on cwd basename — all same-named packages/worktrees collide onto one cache, forcing constant re-imports
- [BUG] STOP CHANGING THE TERMINAL OUTPUT WHEN THE AGENT CORRECTS ITSELF
- [BUG] when performing developer reload, the previous conversation should be resumed by default
- [BUG] Custom mcpb is not picked up by Claude in cowork
- [BUG] Telegram plugin MCP server does not auto-reconnect after Mac sleep/wake or network drop — requires manual /mcp
- [Bug Report] Unable to process - no issue details provided
- [BUG] Cowork: No working stop mechanism; deleting task to stop it permanently deleted user files; usage consumed despite malfunction
- [BUG] VS Code Session History panel ignores hook-set sessionTitle (custom-title present in transcript) — regression in 2.1.170
- [Bug] UltraReview failures consume credits without error reporting. Requesting credit refund.
- [Feature Request] Desktop app: messages typed during active agent turn should interrupt or be configurable, not silently queue
- Desktop app CI panel shows 'Install the GitHub App' despite app installed org-wide (repo transferred between owners)
- [Feature Request] Allow suppressing additionalContext feedback display in Stop hooks
- Streamable HTTP client mis-correlates JSON-RPC error body of 405 GET (SSE stream) response to pending tools/list, breaking stateless servers (Linear MCP)
- Scheduled tasks: wake-time catch-up drops tasks (lastRunAt stamped without a session) and spawn-storms concurrent headless runs
- EnterWorktree (name mode) writes core.hooksPath into the shared repo config; ExitWorktree doesn't restore it
- Inconsistency in agent session management between CLI and desktop app
- Skill allowed-tools grants not applied when the model invokes the skill via the Skill tool
- Gate-skip failure class: agent acknowledges execution gates, restates them, then bypasses them (Opus 3.5 → Fable 5)
- [Bug] Message loss on connection reconnect during input
- [FEATURE] `claude session delete <session-id>` — per-session deletion from server-side storage
- Claude Desktop: add shell configuration support (env.SHELL or per-server shell override)
- [Bug] Unable to use model due to overly restrictive safety guardrails
- [Bug] Model fallback to Opus incorrectly triggered for non-biology domain
- [Bug] Anthropic API Error: Safety measures triggered, automatic model downgrade to Opus
- Add Bookmark/Pin Feature for Long Conversation Navigation
- [BUG] - Claude 1.11847.5 (9692f0) 2026-06-09T17:34:27.000Z prompt window via GUI locks up on a long thread
- [BUG] Settings writer mis-resolves relative symlinked `~/.claude/settings.json` (readlink + logical-dirname join) — `/effort` fails with ENOENT
- [FEATURE] Make the "Virtualization is not available" banner in Cowork dismissible / add an opt-out for users who intentionally run without the sandbox VM
- [Feature Request] Expose modelPicker:saveAsDefault action for keybinding customization
- [Claude Code on the web] Session container restores to a STALE commit (predating current origin tip), causing edits/commits on a stale tree
- [BUG] ArchicadTapir MCP server consistently fails to connect on startup with "Could not attach to MCP server ArchicadTapir"
- [Bug] Anthropic API Error: Message flagged for cybersecurity content in legitimate security project
- [Feature Request] Support security audit capability with Fable integration
- [BUG] Fullscreen (no-flicker/alt-screen) renderer duplicates content lines on wrapped rows (macOS) — classic renderer is clean, A/B reproduced
- Fable 5 overly safe about biology homework problems
- High startup context consumption from system prompt components (~28% before first user message)
- [Bug] Incorrect content categorization or filtering applied to novel application
- [Bug] Overly restrictive content policy blocks legitimate security review requests
- [FEATURE] Native Windows toast notifications (a `windows_toast` channel for `preferredNotifChannel`)
- All subagent (Agent/Task) tool calls fail with `[Tool result missing due to internal error]` while main-thread tool calls succeed
- [Bug] Desktop: Pasting hundreds of files causes app hang; missing confirmation and virtualized rendering
- [BUG] dispatch stuck on Asleep cant reach desktop from IOS
- [Bug] Desktop app session index not updated when .jsonl files restored to disk
- I need the bug report content to generate an appropriate GitHub issue title. Please provide the bug report details.
- [Bug] Unable to generate issue title: Empty bug report provided
- [Bug] Anthropic API Error: Server temporarily limiting requests (rate limit)
- Feature request: `claude://` URL scheme for deep linking into Claude Code with pre-loaded context
- [FEATURE] Opt-in setting to show session cost/usage summary on exit
- [Bug] Assistant text before tool calls not rendering in interactive mode with Fable 5
- [BUG] Platform reported as "darwin" / Shell "zsh" under Git Bash (MSYS) on Windows 11
- Fable 5 false security flag
- Slack MCP: bot messages using legacy attachments return empty body
- Bash tool: foreground stdin is a held-open pipe — stdin-reading commands hang indefinitely and the timeout never fires (Windows/MSYS)
- Tool calls approved via Remote Control (mobile app) are lost: session hangs, '[Tool result missing due to internal error]'
- [BUG] MCP CONNECTOR
- [Feature Request] Claude Fable 5 lacks detailed output and live tool testing compared to Opus
- [Bug] Anthropic API Error: Usage Policy Violation blocking all requests
- Bash tool results silently lost — agent waits forever (no timeout fires); correlates with Remote Control sessions
- [DOCS] Sub-agents page still says subagents cannot spawn other subagents; v2.1.172 allows up to 5 levels of nesting
- [DOCS] Bedrock setup page claims `AWS_REGION` is required and `~/.aws` is not read; v2.1.172 reads region from `~/.aws` when `AWS_REGION` is unset
- [DOCS] Remote Control page still describes footer indicator as "Remote Control active"; v2.1.172 shortens it to "/rc active"
- [DOCS] Model config page says `opusplan` does not get the automatic 1M upgrade; v2.1.172 makes `opusplan` ship with 1M context in plan mode
- [DOCS] Permissions page does not document `WebFetch(domain:*.example.com)` wildcard subdomain matching syntax
- [DOCS] Monitoring-usage page does not list the `model` attribute on `claude_code.lines_of_code.count` added in v2.1.172
- Safety-classifier model switch (Fable 5 → Opus 4.8) fires on benign content and can't be overridden with /model
- [Bug] Claude incorrectly flagging normal coding prompts as restricted topics
- [Bug] Anthropic API Error: Socket connection closed unexpectedly
- [FEATURE] Desktop app: cache and show last 3-4 working directories in the new-session folder picker
- [Bug] Anthropic API Error: Incorrect Security Content Flagging as Cybersecurity Threat
- [BUG] Agent skills: frontmatter field does not inject skill content at startup
- [BUG] cct = TUI: rendered-text character-level garbling with cold-boot persistence
- [BUG] cct = TUI: multi-paragraph live-render-duplication
- [BUG] cct = TUI: render-duplication in turns containing Bash tool_use calls
- [BUG] Latest version of Claude Code on Windows desktop errors on every prompt submit - aka does not work (on my version of windows) - Bun 1.3.14 "Internal assertion failure" (ntdll) crashes BOTH desktop-bundled 2.1.170 and client CLI
- [Bug] Model auto-switching to Opus triggered incorrectly on lambda function review
- [Bug] VSCode: exiting plan mode ignores permissions.defaultMode: "auto" and silently lands in acceptEdits (no choice offered)
- MCP OAuth with pre-configured oauth.clientId still attempts DCR: 'Incompatible auth server: does not support dynamic client registration' (v2.1.172, Box remote MCP)
- [Bug] Auto-Compact triggers prematurely on Opus with 1M token context window
- [Bug] CVP authorization not respected, automatically reverts to Opus 4.8
- [BUG] screen shot with no text give an error and freezes server connection session
- VS Code extension: session listing across git worktrees pulls unrelated repos into Source Control ("Close Repository" undone within ~1s)
- [BUG] 2Pc same account, connector from pc B being proxied through pc A
- [FEATURE] Directly execute a tool with `/tool`
- [BUG] Cowork VM connection timeout on Microsoft Surface Laptop Snapdragon ARM64
- [Bug Report] Unable to generate issue title - no bug report provided
- [Bug] Text block silently dropped from render and transcript after no-comment AskUserQuestion answer
- Headless mode ignores explicit permissions.allow rules for .claude/ subpaths — denied as "sensitive file" with no operator override
- Self-reported: Claude Code violations caught by user 73% of the time (user-configured accountability system)
- Unable to generate issue title - no bug report provided
- Desktop Code tab: model picker shows Fable 5 (1M) while session runs Sonnet 4.6 (200k); context meter pinned at 100% red and exceeding its own denominator
- [Feature Request] Allow reordering of working directory groups in managed agents view
- [Bug] Security flagging triggered for legitimate diagnostic operations
- [Feature Request] Add feedback submission from managed agents view without creating new chat
- [Bug Report] Unable to process empty bug report
- [Feature Request] Support for hardened container image specifications and secure deployment configurations
- [META] cct TUI render-anomaly defect cluster (omission + duplication + corruption families)
Resolved issues (187)
- [Bug] Security scanner falsely flags user code during development
- [BUG] "--remote" web sessions never connect to GitHub remote repo
- PR monitor chips never retire — survive PR close and merge, session archive, and app restart; no way to clear them
- Newly created ~/.claude/settings.json is ignored until restart — permission rules silently inert
- [BUG] Claude Code (Opus 4.8 / 2.1.162) treats an unanswered A/B choice as approval, autonomously launches a multi-agent Workflow, and burns user tokens (recurs despite closed #24318/#40593/#41707)
- [BUG] Remote-control image upload silently dropped — image never reaches model, no error (Windows host, v2.1.162)
- [Bug] Tool trigger sensitivity requires tuning
- [BUG] VS Code extension requires re-authentication every ~8 hours — OAuth token refresh blocked by Cloudflare
- I appreciate your concern, but I think there may be some confusion. I'm Claude Code, an AI assistant designed to help with GitHub issues for the Claude Code CLI tool. Your message appears to be about: - A CVP (Controlled Vendor Programme or similar securi
- Desktop app: Add Auto mode (bypass all permission prompts) equivalent to CLI's --dangerously-skip-permissions
- Phantom user messages: assistant acts on instructions absent from session log (.jsonl) — v2.1.169, embedded terminal
- [FEATURE] Surface detailed session usage stats in VS Code / Cursor extension
- Closed - filed in error
- [FEATURE] Run /rename immediately in the background instead of queuing behind the active turn
- [BUG] Claude Code Desktop: Critical Knowledge Gaps Causing Token Waste
- [BUG] VS Code extension (Windows): Enter/Delete unresponsive in agents manager after returning from an agent (v2.1.168)
- [BUG] Turns frequently fail to render output: leaked "call" text, "tool call could not be parsed", or empty response despite token usage
- [BUG] Cowork (Windows): bash sandbox can't enumerate/delete files on CBFS-based virtual drive, though host-side Read/Write/Edit works
- [BUG] Abnormal Token Consumption Rate in Claude Code - Pro Plan
- User-level skills appear in slash-command autocomplete once per enabled plugin (N plugins = N duplicates)
- Add language selector for built-in voice dictation
- [Feature Request] Allow selection of Claude model versions (e.g., claude-opus-4.6)
- [Bug] File reference autocomplete shows packages instead of project files
- [BUG] can't install claude windows desktop
- Feature request: syntax highlighting in permission approval prompts
- 529 errors stopping me working
- [Bug] Excessive policy violation errors on standard operations
- [BUG] False-positive Usage Policy block on benign field-mapping task (healthcare app)
- [BUG] "Pull request status couldn't be checked" warning appears permanently even with no open PRs
- [Bug] Auto-update warning lacks actionable diagnostics from slash doctor command
- [BUG] Session visible on desktop but not syncing to iPhone Claude app
- [MODEL] Opus 4.8 (Claude Code) still rationalizes skipping process steps ("it's just X" / "my tests are green") -- self-justification defeats explicit standing rules
- [MODEL] Opus 4.8 (Claude Code) still skips user-defined multi-step workflows -- jumps straight to coding without the required plan -> review -> test -> ship gates
- [Bug] Skill(claude-api) returns excessive context in single tool result
- [BUG] Session survey option 3 should be 'Well' not 'Good' — 'doing' takes an adverb
- Allow-listed Read/Write silently soft-deny after ~600s permission timeout in remote/multi-session VS Code
- [BUG] Opus 4.8 drops known constraints during execution (regression vs ~1 week ago)
- [BUG] Severe latency on Sonnet 4.6 in VS Code extension despite green status page (Jun 6, 2026)
- [Bug] Anthropic API Error: Server rate limiting on valid requests
- [Bug] High-frequency tool invocation prefix fragments causing deadlock
- [BUG] MCP OAuth: token exchange silently fails for Robinhood MCP servers -- empty accessToken persisted, empty error from complete_authentication (manual exchange with identical params succeeds)
- [BUG] MSIX packaging causes two critical regressions: ERR_QUIC_PROTOCOL_ERROR blank screen and CoworkVMService ACL lockout
- [Bug] Command autocomplete shows duplicate suggestions, obscuring cursor position
- /btw silently discards conversation — destroying the user's work is the wrong default
- [FEATURE] LaTeX math rendering ($$...$$) is unstable in VSCode extension chat window
- Advisor doesn't work with proxy keys and ANTHROPIC_BASE_URL
- [BUG] GitHub username jamesmitchell720: backend repository index reset
- Support attaching images from iOS Photos in Claude Code mobile app (remote control)
- Permission prompts not rendered while transcript viewer (Ctrl+O) is open
- Claude Code on Desktop app (mac os): Git panel diffs against origin/main without disclosing the base or ahead/behind state — clean working tree looks full of phantom changes
- [BUG] Cowork desktop sandbox — bash mount serves stale/truncated host-edited files.
- Claude ignored memory/history and broke correct code, then validated it with subagents using wrong criteria
- [Feature Request] Anthropic API safety filter incorrectly flags routine system administration commands
- [BUG] Workflow-tool agent() subagents omit x-claude-code-agent-id / parent-agent-id (Task subagents are tagged)
- BILLING] Cannot purchase API credits - card declined at Anthropic layer despite bank and Stripe confirming no issues (Philippines, BDO Amex)
- Chat message silently dropped when switching desktop -> phone client during 'preflight' send phase
- [Bug] Tool invocation fails with string output interference
- [Bug] Claude 3.5 Opus Response Quality Degradation in Extended Sessions
- I'm ready to help generate a technical GitHub issue title for Claude Code. However, I don't see a bug report in your message. Could you please provide the bug report details that you'd like me to create a title for?
- [BUG] Claude Max 5x gift subscription automatically canceled — account reverted to Free plan
- [BUG] No desktop notification for claude ai
- [Feature Request] Add option to disable prompt autosuggestions per skill
- [BUG] Excessive token usage (~20k–30k tokens) for trivial prompts in Claude Code CLI
- [Bug] Unexpected API usage spike without active requests
- [Bug] Model enters infinite loop without reconsidering creative choices
- I need a bug report to generate a GitHub issue title. Please provide details about the problem you've encountered with Claude Code, including: - What were you trying to do? - What error or unexpected behavior occurred? - Any error messages or logs?
- [MODEL]
- Weekly usage shows 0% after Max x20 upgrade but chats remain blocked
- [BUG] Claude Code VSCode extension webview "Not responding" on Windows 11 - terminal works, GUI does not
- Hook subsystem: session metrics in payloads + warm-reset primitive
- [Bug] Command autocomplete not executing on Enter key press
- Native memory leak causing extreme RAM consumption (534GB/hour growth rate)
- [BUG] Startup banner ("What's new") disappears immediately — no time to read release notes
- [BUG] Claude Code silently uses project ANTHROPIC_API_KEY from .env instead of Max plan OAuth, causing unexpected API charges
- [Bug] Plan-approval auto mode rejected with inconsistent gating, disables shift+tab auto
- [BUG] Welcome screen disappears instantly on startup
- [BUG] mcp headersHelper not re-invoked when access token expires on long-lived HTTP transport
- Bring Back /buddy: 511 Reasons Why
- [FEATURE] Claude Cowork 3P’s managedMcpServers should support glob-syntax for toolPolicy
- [BUG] Edit/Write tools fail with EPERM on Windows drive root paths (e.g. D:\)
- Feature request: PostCompact hook event
- [BUG] Terminal graphics residue after context compaction
- [BUG] Dispatch stuck on "Asleep" / "Can't reach your desktop" — all local troubleshooting exhausted
- Worktree naming: avoid person surnames (scientists/heroes); prefer neutral words like fruits
- [BUG] Billing: "Stripe is temporarily unavailable" error on Pro upgrade — multiple devices, multiple days
- [BUG] Cowork runtime missing from macOS app bundle — "Download failed" error on every launch
- [BUG] Cowork runtime missing from macOS app bundle — tab not appearing (v1.1.7714)
- Feature request: user-configurable cache breakpoint hierarchy in settings.json
- Claude Code ignores standing rule to delegate lightweight ops to Haiku subagent — $110+ overcharge in a single session due to cascading context bloat
- [BUG] Cowork slash commands return "Unknown skill" — affects custom org plugin skills AND built-in /feedback
- [BUG] Every command retries up to 10 times ("Retrying in Xs · attempt X/10") and never gets a response. Happens on every single input.
- [BUG] Subagent (Task tool) output silently truncated when final block is a tool_use — regression from #20190 persists in v2.1.139
- Feature request: skill catalog injected once per session, not after every tool result
- [BUG] Cowork live artifact: connector tool calls on page load trigger one-shot Allow/Deny prompt with no "Always allow," even when connector permission is set to "Always Allow"
- [BUG] Shell snapshot captures compdef without completion state, breaking zsh functions that register completions
- Windows: file links in webview chat are not clickable (px() regex breaks on drive letter colon)
- /mcp reconnect should refresh both local and remote (cloud routine) connectors
- [FEATURE] per-plugin `lastUpdated` field in marketplace.json
- Sub-agent returning "Tool result missing due to internal error" silently hangs parent — no error propagation
- [BUG] Where did my money go
- [BUG] Cowork HTTP MCP plugins: silent connection drops, Authorization header dropped on tools/call, and GitHub-disconnect "fix" wipes plugins for the whole workspace
- [BUG]
- [Bug] Agent Teams Planning mode documentation inaccurate for agent behavior
- [FEATURE] Support interactive REPL/console sessions (e.g. bin/rails c, python, node)
- [Bug] Custom command templates insert unintended content in PR descriptions
- [BUG] Sandboxing doesnt work with directories named 'config'
- [Bug] Claude Fable 5 model fallback due to overly aggressive safety filtering
- [DOCS] missing: CLAUDE_CODE_ATTRIBUTION_HEADER - prompt cache with ANTHROPIC_BASE_URL
- [BUG] /usage heatmap misplaces usage by one weekday row; contradicts "Most active day" stat
- False-positive AUP refusal (stop_reason: refusal) permanently poisons session after grepping my own channel transcript
- [BUG] Undefined slash commands are silently sent to the model as plain text — no feedback, inconsistent with malformed-command handling
- [BUG] Deprecation Warning shown despite Model 4.6
- [BUG] Lone surrogate in Bash output produces 'unexpected end of hex escape' 400 — distinct from \u surrogate errors
- `claude agents`: allow specifying a different model per background session at dispatch time
- [UX] "Credit balance" (API console) and "Current balance" (subscription) use identical framing for unrelated billing systems
- [BUG] [Cowork] Windows MSIX: VM bundle and claude-code-vm SDK provision to %APPDATA%\Claude but VM/VirtioFS reads the package-virtualized LocalCache path — "VHDX file not found" then "SDK not verified"
- Slash command run during in-flight server-tool (advisor) call orphans tool_result and permanently wedges session (400)
- Hook runner spawns with deleted session cwd — every hook fails with `ENOENT: posix_spawn '/bin/sh'` and PreToolUse deny-hooks fail open (distinct from the Bash-tool cwd validation)
- [Bug Report] Unable to generate issue title - no bug report provided
- [Bug] GitHub App Installation Status Not Recognized After Reinstall
- [FEATURE]
- Fable 5 overly safe about biology homework problems
- High startup context consumption from system prompt components (~28% before first user message)
- [Bug] Assistant text before tool calls not rendering in interactive mode with Fable 5
- Bash permission patterns do not support ** glob — impossible to scope by both path depth and file extension
- [Bug] Model auto-switching to Opus triggered incorrectly on lambda function review
- [BUG] Claude Desktop installation fails with HRESULT 0x80073CFF on Windows 11 Home
- [BUG] [Windows] Claude Desktop: MSIX auto-update force-restarts the tray-resident app and opens the main window unprompted
- [BUG] Dispatch permanently offline - same as #45937
- [BUG] Network error
- [Bug] Unable to access claude.ai settings on macOS and iOS - billing update blocked
- [BUG] Dispatch permanently stuck "asleep" on Windows 11 — server-side session not recovering
- [BUG] Authentication Loop - Apple Private Relay <> Claude Desktop
- [Bug] Agent unfocused reasoning dilutes code quality and introduces bugs
- [Bug Report] Unable to process request - missing bug report details
- [Bug Report] Unable to determine issue from empty report
- [Bug] Claude Code Desktop App Code tab crashes on startup with exit code 1 after macOS user migration
- [BUG] Custom theme "custom:<slug>" in settings.json not applied at session startup (v2.1.132) — falls back to auto
- /voice: second concurrent session fails with "No audio detected from microphone"
- Show context/token usage indicator in VSCode extension's input bar
- Feature: Better session organization (folders/projects/tags) like Claude.ai
- [BUG] Sandbox Config tab filesystem restrictions dropping leading slash on WSL2
- [BUG] Voice "hold" mode causes claude to be suspended (SIGTTIN) when terminal window loses focus
- [BUG] Claude hallucinates Claude Code's own settings/hooks schema
- [BUG] Silent data loss: Cowork Edit/Write tools truncate long non-ASCII content; Read tool serves stale cache that hides it
- [FEATURE] Unable to switch default terminal on Windows
- [BUG] Session usage counter shows 38% used with no user activity after login
- Model silently substitutes a different approach than the one specified, without surfacing the deviation
- [FEATURE] Add settings.json key to default worktree-exit prompt to "keep"
- [BUG] Claude Code on the web silently stops mid-task after tool calls
- [BUG] Model silently stops mid-task after tool call without generating follow-up
- [BUG] Ultrareview: failed runs reported as completed to agent
- [BUG] Claude stops responding mid-task after successful tool calls
- [BUG] Claude desktop app blocks new sessions with "Git is required" despite git being fully functional
- When a stale-training assumption proves wrong, Claude patches forward instead of re-grounding — the cascade isn't course-corrected
- [BUG] Bash tool sends SIGTERM instead of SIGINT when cancelling commands, corrupting build state
- [Bug] Claude Code performance degradation - significantly slower execution times
- Remote control commands not delivered while /status screen is open (flush on close)
- Claude under-uses the persistent auto-memory store during long multi-session work — user re-explains context every session despite the mechanism existing
- Agent biases toward 'act first' on troubleshooting tasks even after explicit user instruction to search first
- [BUG] PowerShell encoding bug and that you had to spend credits to fix a problem
- Monitor details panel should display the `description` field
- [Bug] Historical sessions not displaying messages
- [BUG itHub repositories not visible in Claude Code repo picker despite app being correctly installed**
- [Feature Request] Chrome extension conversations not saved to Claude chat history and lack persistent memory across sessions
- [Bug] Anthropic API Error: Output blocked by content filtering policy
- [MODEL] Unprompted branch opened by Claude Code
- [BUG] Claude Code "No Response Requested" Loop
- Permission prompt appears despite --dangerously-skip-permissions flag
- Cowork: `Read` fails with misleading `EINTR` on user-owned files created outside the Cowork session (correlates with `com.apple.macl` absence on macOS)
- [Bug] Auto-compact triggers far below 1M window on Opus 4.7 [1M] — severe token burn regression on Max 5x (post-2.1.116)
- Toolbar icons overlap and spacing breaks on Arabic Windows environment
- [BUG/feature] Add Windows-aware concurrency cap for parallel Agent dispatches (msys fork corruption)
- [BUG] Claude Code's system reminders use the same pattern as prompt injection
- Claude-in-Chrome MCP bridge fails on Windows: framing protocol mismatch between MCP client and native host
- [Bug] v2.1.121 Windows: Claude Code REPL silently exits inside PowerShell host during long-running agent + dense Bash subprocess chain
- Support renaming the active worktree from within a Claude Code session
- Sycophancy and completion bias causing measurable financial harm to professional users
- [BUG] Inconsistent behavior when conflicting saving policies exist across CLAUDE.md memory levels
- [BUG] Pattern of undisclosed 1M context billing changes affecting Max plan users — cross-issue summary
- Auto-mode classifier blocks authorized operator workflows
- [BUG] Read tool downsamples PNG based on file size, not pixel area — same image at smaller bytes arrives readable
- WebFetch summarizer fabricates <system-reminder> blocks indistinguishable from real harness reminders
- [Feature Request] Refine safety classifier to reduce false positives on legitimate scientific computing
- [BUG] Cowork VM connection timeout on Microsoft Surface Laptop Snapdragon ARM64
- [BUG] SSH remote connection fails with 'failed to upload file: no such file' when Windows username contains spaces
- [BUG] Claude Code unresponsive over Remote-SSH on a Linux server under memory load (both VS Code extension and CLI)