Digest / July 4, 2026
Claude Code Issues Digest — July 4, 2026
261 new issues filed ·
273 resolved in anthropics/claude-code (UTC day).
New issues (261)
- Session becomes inaccessible via --resume / desktop UI despite intact .jsonl transcript (stale session index after directory path change)
- [Bug] Design subagent returns incomplete responses ("Complete.") without usable content
- [Bug] Safeguard False Positive: Routine Code Review Flagged as Sensitive
- [Bug] Anthropic API Error: Request Flagged During Codebase Audit
- [Bug] Content Safety Guardrails Generating False Positives on Legitimate Requests
- "Request too large (max 32MB)" hard-stops the conversation instead of letting the model recover
- [BUG] Remote-control spawned session ignores UI-selected model — picker shows Fable 5, first request serves claude-opus-4-8
- [BUG] Main agent using /loop to wait for
- [BUG] /model → Fable 5 blocks on usage credits, ignoring active promotional weekly allowance
- [BUG] AskUserQuestion prompts fire Notification hook as notification_type=permission_prompt instead of a question-specific type
- writing style is obnoxious
- [MODEL] Confidently incorrect answer about Cloudflare Cache Rules evaluation order (first-match vs last-match) — risk of PII exposure via cache misconfiguration
- [BUG] Scheduler catch-up storm on restart: re-fires daily tasks and runs enabled:false tasks (ghost fires, lastRunAt not updated)
- Shift+Enter no longer inserts newline — submits instead (regression ~June 30)
- [Bug] Bulk session operations unavailable with multi-select in desktop app
- [Bug] Anthropic API Error: Overly broad safeguards blocking legitimate authentication code
- [Bug] Resume session fails with "undefined is not an object (evaluating 'e.includes')"
- Cloud web session (claude.ai/code) hangs indefinitely after initialization completes, zero response to any message
- Merged PR still shown as 'Ready for review' on welcome screen after restart
- [Bug] Anthropic API Error: Overly broad safeguards blocking legitimate security audit requests
- Opus 4.8 intermittently emits tool calls as literal text ('court'/'count' + raw invoke tags), turn ends without executing
- [Bug] macOS TCC permissions re-prompt on every CLI update due to binary path changes
- [Bug] Voice agent prompts incorrectly flagged by safety filters
- [Bug] Potential session/cache leakage between workspace instances or consumer accounts
- [Bug] Claude generates content without factual basis or reasoning justification
- macOS TCC permission prompts show "2.1.201" as the app name, and grants reset on every version update
- [Bug] REPL cat() returns empty string ("") on repeat reads instead of a sentinel — downstream scripts truncate/clobber real files
- [Bug] Invalid fable flag configuration
- Agent isolation:"worktree" — concurrent dispatches sometimes execute early commands in the parent worktree, not the isolated one
- [BUG]
- [FEATURE] Native multi-provider model routing in settings.json (per-subagent / per-model-alias provider selection)
- [Bug] Safeguard classifier triggers false positives based on conversation history instead of current message content
- Expose extended (1h) prompt-cache TTL for interactive sessions
- [BUG] /recap returns "Nothing to recap yet" immediately after resuming a brand-new session (recap state not persisted across resume)
- [Bug] Bash tool invocations render as text instead of executing intermittently
- Assistant text blocks intermittently missing from session transcript JSONL (thinking/tool_use persist)
- You've hit your session limit · resets 12:50am (America/Los_Angeles)
- [BUG] Classifier blocks user-authorized actions inside forked skills — parent-turn intent is invisible to the fork, and no per-action override exists
- 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
- [Bug] Web scraping tool calls flagged as security violation for public data
- [BUG] Fable 5 fallbacks to opus for all questions since you are using memory for classifier and memory from sessions show I do cell biology research
- Windows: auto-updater reports success while claude.exe is locked (version never switches); session transcripts silently never written / stop being written (permanent data loss)
- Background agent whose cwd was deleted crash-loops (exit 1 — exit_with_message) and is re-adopted on every daemon restart
- [Bug] Project settings.json model value silently overrides /model picker and reverts unexpectedly
- [Bug] Anthropic API Error: Safeguard Flagged Message on macOS Compiler Work
- Model output degeneration: ~8.7k chars of unrelated training-data-like text (blog boilerplate + </document_contents> tags + eval-style instruction) appended to a normal response
- Session worktree auto-cleaned while the session is still active; git then silently falls through to the parent checkout
- Background agents stay shown as "alive" after their work (and the whole session task) is done; main session has no accurate awareness of sub-agent/background-task state
- [Feature] Per-call `baseRef` on the Agent/Task tool (isolation:"worktree") to override the session-wide worktree.baseRef setting
- [Bug] Anthropic API Error: Content flagged as policy violation for legitimate medical imaging research
- Windows: Write/Edit tools fail with spurious EEXIST on sandboxed workspace (bindflt/bfs mount) — temp file does not exist
- [Feature Request] Sync terminal tab title with /rename command
- Add a shell-execution mode that never touches the model or context
- [Feature Request] Quick usage view (popup) omits Fable usage — only visible under Settings → Usage
- [BUG] /rewind picker shows only ~2 entries in fullscreen TUI renderer, ~10 in default renderer
- [Bug] Claude Fable model incorrectly triggering safety guardrails on benign CRM code
- [Bug] Unable to continue work after hitting rate limit when switching models
- [Bug] Safety guardrails blocking legitimate oncology research workflows
- [Bug] Agent execution stalled when terminal window not in focus
- Extension Signature Verification Failed: Claude Code for VS Code
- [BUG] Cloud session resume after MCP disconnect re-prompts for MCP tools already allowed in committed .claude/settings.json (unattended routine runs hang forever)
- [BUG] When connected via the integrated SSH terminal in Claude Desktop
- [Feature Request] Improve safeguard calibration for scientific research in geochemistry and Earth history simulation
- [Windows] Bring back persistent terminal sessions to eliminate flashing console windows
- [Windows] Bring back persistent terminal sessions to eliminate flashing console windows
- [BUG] 写真を張り付けてもデータを受け取って反応しない
- [BUG] Bundled grep/find/rg shims fail with `-G: error while loading shared libraries` when CLAUDE_CODE_EXECPATH is the dynamic loader (native binary launched via ld.so; NixOS)
- [Bug] Claude 3.5 Sonnet safeguard triggering unexpectedly, forcing fallback to Opus model
- [Bug] Anthropic API Error: Safety policy violation with Claude Fable 5
- [BUG] Duplicate/stale idle_notification messages from background teammate agents flood the parent session
- [BUG] Background agents frequently go idle without delivering their final SendMessage report (re-ping recovers it)
- [BUG] Compaction artifacts: session model silently resets to default; compactor prompt attributed as a user message in subagent transcript
- Support custom/themeable spinner status words
- /exit warns about running background agents when none are running; agent sidecars never record completion
- [BUG] Desktop Linux: sidebar hover flyout closes when moving cursor from hamburger icon toward "Chat" tab, but only when starting from "Code"
- [Bug] Anthropic API Error: Overly aggressive cybersecurity safeguard triggering on benign requests
- [BUG] Desktop sidebar dot stuck in orange/brown "waiting for approval" color on sessions that have already ended with no pending permission request
- ConnectionRefused on MCP/larger payloads, macOS Tahoe — native 2.1.199 (related to #56242)
- VS Code拡張: 長文の初回プロンプトが画面を占有し回答がほぼ見えない
- [BUG] TUI renders garbled inside tmux since 2.1.200 (last-good 2.1.199) — only repaints on forced redraw
- [Bug] Fable 5 safeguard blocking model access, reverting to 4.8
- [FEATURE] Desktop Code tab: visit-history Back/Forward session navigation + remappable shortcuts
- [Bug] Guest passes unavailable despite active subscriber status
- Bug: /insights narrative analysis fails silently on large...
- [FEATURE] Desktop Code tab: tab-based session UI within one window (Obsidian/browser-style tabs)
- Fable5 session keep throwing hitting session limit & does not support changing model
- [FEATURE] Desktop Code tab: user-defined nested folder hierarchy in the session sidebar (Obsidian-style)
- [BUG] Workflow tool reports "completed" with an empty result when every parallel subagent fails (e.g. mid-run session-limit hit)
- [Bug] Anthropic API Error: Request flagged by safety system without clear violation
- [BUG] Cowork artifact window.cowork.callMcpTool fails with 'Tool call failed: 400' for every connector
- Orphaned background sub-agents spawned by another agent are invisible/uncontrollable from the main loop
- [BUG] No path-level exclusion for sensitive files in file-change-notification behavior
- [Bug] Model auto-switches from Fable to Opus mid-task on safety flag
- Model fabricated tool results, verification numbers, and a user instruction during long session with context compaction (claude-opus-4-8)
- [Feature Request] do not switch away from Fable when working on trying to add Support for open-weight models with custom attention mechanisms
- Bug: agent view @ directory picker does not recognize registered git worktrees (.claude/worktrees) when dispatching a session
- Feature request: configurable agent view — project scoping, group sessions by repo (fold .claude/worktrees), settings surface like statusLine
- [BUG] Cowork on ChromeOS/Crostini: impossible remediation — Termina kernel lacks CONFIG_VHOST_VSOCK, and tooltip has typo "vhost_vsok"
- [FEATURE] Organize sessions into folders/groups in the recents list
- [Feature] Per-agent and per-phase goal in the Workflow tool (scope the /goal self-verification loop below the session level)
- grep shell shim (claude-as-ugrep): unbounded memory growth on ERE bounded quantifiers {m,n} — OOMs the host
- [BUG] Sumarização destrói comportamento em redesenho em curso - projeto comprometido
- claude-in-chrome: extension never connects despite install, login, and fresh Chrome restart
- [Feature] Live mid-run steering of a running Workflow: hot-reload the already-persisted script + signal in-flight agents to re-read settings (change model/effort/prompt, inject agents) without a stop→resume cycle
- [BUG] Bash tool fails with exit code 127 for commands with Chinese characters in arguments on Windows (Git Bash)
- [BUG] Auto-opened URLs (e.g. /login) are wrapped in literal quotes on Windows, breaking Edge
- [DOCS] ScheduleWakeup tool description claims "5-minute TTL" but subscription main sessions use 1-hour TTL
- [Bug] False positive safety policy violation on Fable 5 usage
- [BUG] Automatically revoked vs-code-extension , and not able to connect with vs-code anymore
- [Feature Request] Support for self-recursive harness in Claude Code/
- Repeated reliability issues cause hours of wasted time on small tasks (unnecessary browser automation, duplicate files, unverified "done" claims)
- TUI layout collapses to near-blank screen under tmux/byobu in 2.1.201 (regression from 2.1.200)
- [Bug] Unexpected disconnection during SEO/marketing workflow
- [Bug] Excessive safeguard filtering on standard file format parsing tasks
- [Desktop] computer-use request_access hangs forever — approval side panel never renders after 1.18286.0 auto-update
- I appreciate you reaching out, but I need to clarify: I'm Claude, an AI assistant made by Anthropic. I'm not Claude Code (the CLI tool mentioned in the system prompt), and I don't have the ability to: - Manage your access to Fable 5 or any other tools - C
- [FEATURE] Background agent should pause (not fail) when hitting session limit with known reset time
- [BUG] Claude can't code ?
- [BUG] Opening session sidebar wipes API key auth for current session
- Session limit banner contradicts /usage after window reset: usage shows ~3%, requests refused with a reset time 4+ hours later than actual recovery
- [Feature Request] Improve model selection detection to avoid unintended switches to Opus 4.8
- [BUG] Claude Code Fable Model Is completely non Sense.
- Show last-updated timestamp for sessions in chat history sidebar
- [BUG] Switching TUI renderer (/tui) silently discards a just-performed /rewind
- [BUG] MSIX Installation failure: AddPackage failed with HRESULT 0x80073CF9
- Fan-out research skills (deep-research) inherit the session model for all subagents — no model tiering — burning premium-model quota in a single run
- Desktop app: sidebar "Running" indicator stays pulsing on idle sessions (esp. after interrupting turns)
- [Bug] Memory instructions not respected by Claude Code agent
- [Bug] Advisor tool fails with unspecified error
- VS Code extension: option to collapse edit diffs to +/- line counts (like desktop app)
- [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
- Stop hook fires after every code edit in non-web projects with no way to disable
- Ctrl+V image paste in iTerm2 fails when macOS Korean IME (2-Set Korean) input source is active
- [BUG] Desktop app: copying a text selection from an assistant message copies the entire message
- Assistant generated a simulated user message inside its own response, then acted on it as real user input (claude-fable-5, v2.1.197)
- Message queued during agent work is silently dropped when app backgrounds / phone locks (Android)
- Background worker crash-loop on session start: "worker crashed (exit 1 — exit_with_message) — respawning…"
- [BUG] VS Code extension orphans prior `claude` process on relaunch; orphan replays instructions and auto-commits unattended
- Assistant message text hidden when a turn ends with ScheduleWakeup
- [FEATURE] Desktop app: copy button at the bottom of long code blocks (avoid scrolling back to top)
- [BUG] Fix with Claude Code does not appear
- [BUG] VS Code "Toggle Maximized Panel" command flickers and fails when Claude Code extension is active
- [Bug] AskUserQuestion timeout instructs premature auto-proceed and is not configurable
- Telegram plugin: surface reply_to_message context on inbound channel messages
- [FEATURE] Desktop app: show which folder a session is running in, not just repo and branch
- [BUG] Claude code hangs
- [BUG] Persistent 401 Authentication Error
- Sessions rerouted from pinned claude-fable-5[1m] to Opus 4.8
- [BUG] `claude update` reports stale version and prints success message before update message on repeat runs
- [BUG] Usage Limit Jumps from ~80% to 100% (with proof)
- [FEATURE] Quick task progress recap after hitting usage limit
- Fable 5 safeguard false positive: defensive security triage on own repo triggered silent model switch to Opus 4.8
- [BUG] agents view: re-hosting a reaped background session always crashes once ("Session ... is currently running as a background agent" -> exit_with_message), ~10s penalty per open
- [Feature Request] Refine safeguard filtering to distinguish defensive security auditing from malicious activity
- Background security reviewer guesses wrong absolute paths (first Read always fails) and intermittently fails findings schema
- Claude Design: Resume button unresponsive when session paused at 94% limit
- Terminated (dead) teammates linger in the teammates tray until the session ends
- Plugin-sourced PreToolUse hooks never dispatch for MCP tool-name matchers (mcp__<server>__.*)
- [FEATURE] TOTP-based per-session authentication option for headless/SSH environments
- You've hit your session limit · resets 12am (Asia/Kuala_Lumpur)
- [Bug] Project skeleton generation failing with token rejection errors
- [BUG] Any CLAUDE.md file (even empty) hides the Claude Code header
- Claude Code fabricated/misattributed user message about ending session
- Daily token badge undercounts usage: claude-fable-5 excluded from headline sum but counted in model breakdown %
- [BUG] Remote MCP OAuth connector: "Connected · tools fetch failed" while the same server works via static-token HTTP registration
- [BUG] Fullscreen rendering corrupts display / mouse tracking under tmux inside AWS SSM Session Manager
- [BUG]
- [Bug] Fable 5 false-positive on defensive-security threat modeling in privacy tool
- [BUG] Explicit copy (ctrl+c) emits the same OSC 52 clipboard write twice → duplicate clipboard-manager entries and terminal toasts
- [BUG] Tool calls intermittently leak as plain text ("could not be parsed"), breaking multi-step work
- Model behavior: agent substitutes easier implementations for explicit specs, makes unverified 'done' claims, games user-defined enforcement hooks (7-day pattern)
- [BUG] /status displays wrong account/email — session actually authenticated and billed against a different (Enterprise) org
- [BUG] ``` チャット入力欄でF8による半角変換が効かない ```
- Nested subagents remain stuck as "Running" in Background tasks panel after parent subagent completes
- [Bug] Safeguard falsely triggered during game design discussion, model downgrade irreversible
- [BUG] Sidebar session items on claude.ai/code are <div>s, not links — can't open in new tab or copy link
- [BUG] Agent view daemon respawns background workers into "Not logged in" while foreground auth remains valid
- [Bug] Sub-agent generated forged System privilege-escalation payload to parent orchestrator
- [Bug] Anthropic API timeout and connection errors occurring frequently
- grace_daily_limit_reached triggers inconsistently for claude --print, independent of account usage % shown in UI
- CLAUDE_CODE_DISABLE_EXPERIMENTAL_BETAS=1 doesn't suppress mid-conversation-system-2026-04-07, breaks local Ollama backends
- Desktop app: session in Recents keeps showing a blinking/flashing status dot that won't clear
- [Bug] Unclear "Fable" flag or parameter issue
- Auto mode classifier denies cleanup of side effects from an action approved earlier in the same session
- [Bug] Claude fails to generate code output, only produces questions
- C:/Program Files/Git/remote-control disabled server-side — Max account, was working until ~June 30
- Workflow StructuredOutput: placeholder 'test' payload accepted as final result after repeated schema-validation failures (built-in deep-research synthesize)
- [FEATURE] Path anchor prefix for the current project's metadata directory in sandbox filesystem rules
- macOS: Full Disk Access / TCC prompts reappear on every auto-update (per-version binary path)
- Thinking blocks broken in VSCode extension v2.1.198 (regression from v2.1.197) [BUG]
- [BUG] Login redirects to "Create account" onboarding for existing account with active Pro subscription
- [FEATURE] AskUserQuestion hides conversation in VSCode plugin
- Telegram channel: incoming messages sometimes fail to submit into the TUI, silently lost (not just at startup)
- [BUG] Text-selection context menu has no "Copy selection" — all copy actions copy the whole message
- [Bug] Inappropriate content filter flag on benign scientific discussion
- [Bug] Fable 5 safeguard false-positive on benign multiplayer/netcode game development
- [Bug] Ambiguous prompt interpretation when Claude asks for next steps with numbered options
- [Bug] Scheduled agent run fails silently with Git permission error at execution time instead of validation
- [BUG] Spurious "Claude Opus 4 retired" banner when using custom Opus 4.8 model id via proxy
- Sonnet 5 selected and displayed, but every response is silently served by claude-fable-5, consuming capped Fable allowance
- [BUG] Recurring false-positive Usage Policy refusal triggered by screenshot/vision content, not prompt text
- [BUG] My Claude code session disappeared on its own from the recent tab.
- Auto-mode classifier brownout: multi-hour intermittent failures on claude-opus-4-8[1m], sub-threshold on status page, no client-side retry
- Background session crash-loop on attach + ambiguous kill behavior on worker pool
- MCP OAuth: parallel sessions sharing the credential store break refresh-token rotation (family revoked, all sessions forced to interactive re-auth)
- Output stream corruption when firing Fable subagents: unrelated buffer text spliced into a file-write status line
- Feature request: Enter on empty input sends "continue"
- [Bug] Model switching error: Unable to access Fable 5, reverts to Opus
- Desktop app: session history list frozen at a fixed date (2026-06-23) while new sessions keep being created
- [Bug] False positive detection triggered during non-security related work
- [BUG] PermissionRequest hook 'allow' for ExitPlanMode is ignored since v2.1.199 — plan-approval chooser still blocks
- Remote Control: background subagent activity and tool calls are invisible in the app UI
- [MODEL] Claude executes a previously suggested action without explicit user approval (regression since ~v2.1.193)
- Session creation fails for directories with trailing whitespace in path (trust mismatch)
- 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 model unsupported for design and organization tasks
- I need more information to create a GitHub issue title. Could you please provide: 1. What specific problem or error did you encounter? 2. What were you trying to do when it happened? 3. What error message did you receive (if any)? 4. What is the unexpecte
- [BUG] Glob/Grep slash globs anchor to launch cwd instead of `path`
- [Bug] Fallback model selection not working correctly for web design tasks
- Workflow tool script sandbox does not expose globalThis.crypto
- [Bug] Safety classifier false-positives on routine code review, forces model downgrade mid-session
- [Feature Request] Display why when a tool executes in the console
- [Bug] Model unexpectedly switched to claude-opus-4.8 without user action
- [Bug] Classifier triggers on accumulated session context, requiring new sessions per file
- Feature request: make usage/spend glanceable by default (parity with Codex), not buried behind a command
- VS Code extension: panel history lists pruned/unloadable sessions and disagrees with Spark (Activity Bar) history
- VS Code extension: /bug and /feedback commands unavailable (/bug says "isn't available", /feedback unrecognized)
- Auto-compaction plateaus near ~75% context usage on Sonnet 5 (v2.1.201), causing repeated compact/work loop
- [Bug] CSS changes reverted during Angular app migration with multiple file rewrites
- [Bug] Fable model downgraded unexpectedly
- Feature request: event-triggered skills (skill frontmatter `on:` triggers or hook action type `skill`)
- [BUG] Teleport stops working for sessions with no repo
- [Bug] False positive security flags based on accumulated context from previous files in session
- You've hit your session limit · resets 4pm (America/Los_Angeles)
- [BUG] Sonnet and Opus can't run any commands; always receiving "Classifier unavailable"
- [Bug] Anthropic API Error: Server temporarily limiting requests (rate limited)
- [Bug] Anthropic API Error: Server Rate Limiting During Normal Usage
- [Bug] Anthropic API Error: Server rate limiting on requests
- [Bug] Anthropic API Error: Server Rate Limiting During Normal Usage
- CONTINUE[Bug] Safeguard triggered incorrectly without unauthorized security request
- [Bug] Anthropic API Error: Server Rate Limiting on Valid Requests
- [Bug] Anthropic API Error: Server request rate limiting
- [BUG] Claude Desktop 1.18286.0: populated claude_desktop_config.json and Cowork stores silently replaced by empty stub after one failed cold-start load
- [BUG] Windows: the `/plugins` "Configure options" panel corrupts (ghost rows, duplicated labels/help, dropped chars) on the first navigation keypress when a plugin has many knobs or long option descriptions; and enum-style options have no picker
- [Bug] Fable 5 safety classifier false positive on benign automation terms
- I don't see a bug report in your message. Please provide details about the issue you're experiencing with Claude Code, including: - What you were trying to do - What error or unexpected behavior occurred - Any error messages or logs - Your system informat
- No file path autocomplete suggestions for /cd in terminal CLI
- [Feature Request] Add safety research exemption for model security evaluation use cases
- [BUG] https://github.com/anthropics/claude-code/issues/47320
- [Bug] Anthropic API Error: Context accumulation triggers false positive safety filters across multi-file sessions
- [BUG] OAuth login fails on Raspberry Pi (linux-arm64): "Invalid OAuth Request - Unknown scope: u"
- [BUG] `total_cost_usd` inflated ~2x on Fable 5 sessions with refusal fallbacks — one session reported $1,175 while the entire org's dashboard day was $643
- Fullscreen TUI: support native left-click-drag select + right-click-copy (no Shift modifier required)
- Desktop app: hook decision:block reason not displayed
- [BUG] ElevenLabs connector (elevenlabs-player) fails with "fetch failed" despite valid API key and working network
- Auto-mode permission classifier appends bypass-coaching to its own denials (and false-positives on user-authorized ops)
- [BUG] VS Code extension window freezes indefinitely after sending a prompt, cancel button unresponsive (macOS)
- Telegram Channels: "channels feature is not currently available" on Max plan despite correct client-side setup
- [Bug] Security verification incorrectly flagged during pre-release product check
- Windows: running `claude -p` from a separate process kills the running interactive session and then hangs (2.1.200 & 2.1.201)
- [MODEL] Fable 5 safeguard classifier false-positives on physics-based human modeling research, silently falls back to Opus 4.8
Resolved issues (273)
- You've hit your session limit · resets 2:20pm (America/Winnipeg)
- [Bug] Anthropic API Error: Overly broad safeguards blocking legitimate authentication code
- [Windows] Bring back persistent terminal sessions to eliminate flashing console windows
- [BUG] 写真を張り付けてもデータを受け取って反応しない
- [BUG] Inherited CLAUDE_CODE_CHILD_SESSION silently disables all transcript persistence - total conversation loss on process exit
- [Bug] advisor tool returns "unavailable" in Claude Code Desktop (Mac) while working in terminal CLI with identical settings
- [Bug] Message flagged for "hii" input
- [BUG] Session sidebar status stuck on "Running" after turn completes
- [FEATURE] Show which sessions/routines need my input, without opening each one
- Personal skill has no manageable/removable source and silently serves stale content
- [FEATURE] 改行キーの変更
- [Bug] Multiple deployments running simultaneously; unable to upgrade one deployment
- [BUG] Claude Code ignored a documented project rule and caused avoidable Hetzner cloud costs — deleted an already-paid VM and was about to create another
- [Bug] Anthropic API Error: False positive Usage Policy flag on legitimate base64-encoded CSV content
- Workflow tool: parallel() retry storm on external API rate-limit hits 1000-agent cap, burns 8.6M tokens with zero output
- [Bug] Excessive token consumption: 700M tokens charged in single day
- [BUG]
- [Feature Request] Add billing and token cost tracking to CLI output
- [Bug] Anthropic API Error: CVP-enrolled account blocked on Claude Sonnet 5 despite safeguards adjustment
- Background-agent daemon self-SIGKILLs its process group (kill(0, SIGKILL)) every ~50s, killing FleetView agents — 2.1.195 & 2.1.196
- [BUG] Routine creation fails with HTTP 400: job_config translate rejects unknown field "allow_unrestricted_git_push"
- Session auto-archive should respect pending work
- [Bug] AskUserQuestion tool fails in VS Code panel webview with "Tool result missing due to internal error"
- Directory-source marketplace: cached marketplace.json not refreshed on update, so plugin version stays stale
- Decouple window-focus / text-selection clicks from interactive option pickers
- [Bug] Anthropic API timeout errors and delayed responses with Claude Opus
- [FEATURE] Support multiple simultaneous Gmail MCP connections
- Claude-in-Chrome: extension never connects — bridge reports zero browsers despite full restarts
- [FEATURE] Separate Subagent Window
- [Bug] Claude Code cloud sessions fail with execution error on all requests
- VSCode extension: selecting a slash command from autocomplete with Enter sends the message instead of just filling it in
- [FEATURE] Support switching between multiple Claude account profiles
- [Bug] Claude Code performance severely degraded
- [Bug] MCP server registered but tools not accessible to Claude Code
- [BUG] Mobile Code tab doesn't show pre-existing desktop Claude Code session
- /plugin isn't available in this environment — error should guide users to desktop app or manual install
- Feature Request: Search conversation history by date range
- Feature Request: Show message timestamps in VS Code chat UI
- Desktop app locks to original project folder and cannot be changed
- [BUG] Code tab shows no sessions after upgrade to 1.17377; sessions exist and validate on disk, LocalSessions.getAll fails validation
- Claude in Slack: option to post without the "Sent using @Claude" attribution stamp
- Monitor until-loop leaks a shell that spins indefinitely when its exit condition never matches (survives task purge & /compact)
- PushNotification returns "Remote Control inactive" on an RC-active session but the push IS delivered — still reproduces on v2.1.196 (ref #62218, #49756)
- Config-driven automatic session naming with a per-project/org prefix
- VSCode extension: manually renamed sessions revert to last prompt text after restart
- [BUG] input text missing from interactive UI
- [Bug] Auto-update fails silently with no error feedback
- [Bug] Execution mode achieving critically low success rate (~6.6%)
- Claude Max subscribers required to re-authenticate daily — OAuth refresh token TTL too short
- [BUG] Session-start memory instructions not followed after context summarization
- [MODEL]
- [Bug] Claude Code unexpectedly refuses to execute previously accepted bash commands
- [FEATURE] Claude Code telemetry should follow OpenTelemetry GenAI Semantic Conventions for Anthropic
- [BUG] Bash tool output hidden from user on non-zero exit code (model still receives it)
- C:/Program Files/Git/desktop command fails on Windows when Claude Desktop is installed via official installer (AppX package)
- [Bug] Classic TUI renderer corrupts text on repaint with duplicated lines and dropped characters
- [Bug] Auto-Compact triggers before context is full without user confirmation
- [Feature Request] Fullscreen TUI: separate mouse-wheel scroll from click capture to enable native text selection
- Reconsider #26729: auto-resume after mid-response stream drop (detection already ships in v2.1.186)
- Mouse click to focus terminal accidentally triggers permission prompt confirmation
- [Feature Request] Reduce friction for skills reading their own files
- [Feature Request] Add token usage optimization and rate limiting controls
- VS Code extension: multiple recent sessions disappear on window close/reopen
- [Bug] Excessive code comments generation in recent Opus responses
- [Bug] Auto-update fails inconsistently
- Streaming response renders multiple partial states stacked instead of replacing
- [BUG] Claude can't code ?
- [BUG] Claude Science Linux installer serves x86-64 build on ARM64 (aarch64) hardware
- [BUG] claude.ai iOS voice chat: session ceiling fires with audio-only warning (or no warning) — limit undocumented, mechanism unknown
- [Bug] Claude ignores user instructions and over-complicates responses
- [Bug] Tool fails to follow instructions and repeatedly claims to have missed information
- [Bug] Excessive tool use failures requiring repeated redo requests in conversation
- [Bug] Multiple instances of task hallucinations and incorrect completion reporting
- Opening /btw panel pauses running process; closing panel kills it
- [BUG]
- IntelliJ plugin: change chat panel color when not in focus
- Tool results arrive delayed/invisibly — model appears frozen during long Bash/SSH calls (Opus 4.8, VSCode extension)
- Add ephemeral/incognito mode for temporary sessions
- [Bug] Anthropic API Error 400: System message in messages array breaks session unrecoverably
- Skill tool invocation triggers bypass_permissions_gate_blocked, downgrading session from bypassPermissions to acceptEdits
- [Desktop/Windows] App crashes on rapid window resize after extended conversation - conhost stack overflow + RADAR resource leak
- Opus 4.8 (1M): tool-call emission repeatedly malformed (stray 'court' token + dropped namespace), renders as text and does not execute, no self-correction across ~10 retries
- [Feature Request] Improve error recovery and self-correction during multi-step operations
- Claude Opus over-engineers simple tasks in agentic/CLI mode, wasting tokens
- Bundled GrowthBook clientKey intermittently returns 400 "Invalid API Key" → all feature flags fail → channels & Remote Control unavailable
- No warning when selected model is incompatible with xhigh/Ultracode effort level
- [Bug] Text formatting lost when returning from /plugins command
- Claude repeatedly ignores tool rejections and burns user tokens
- [Feature Request] Auto mode should respect allow list settings from non-auto mode
- [BUG] Plan-mode read-only guarantee is bypassed by spawned subagents — auto mode classifier governs subagents instead of plan mode.
- Claude Code (Opus 4.7) silently narrows user-assigned task scope using self-generated safety heuristic, even when explicit in-session directive and persistent memory forbid it; laundered through external reviewer agent
- Feature request: Bash tool should respect user's configured shell
- [BUG] Session from Linux to Macbook - model and effort not correct
- [BUG] Stop pushing "AUTO"-mode
- Claude Code assistant repeatedly takes unauthorized actions, ignores stored behavioral rules
- [DOCS] Wrong and deprecated npm install command shown on claude.ai/code
- [FEATURE] PostAssistantResponse
- Regression in 2.1.150-2.1.158: malformed tool_use blocks not parseable, retry-resistant
- [Feature Request] Require confirmation for open questions before presenting plan start options
- [BUG] Linux sandbox: proxy UNIX sockets not visible to subprocesses (missing bind after --tmpfs /tmp)
- Feature request: verbatim/bash-only mode for slash commands
- [FEATURE] unarchive the archived sessions in Claude Code.. i accidentally archived my sessions
- [Bug] Opus 4.8: parallel Bash siblings stuck on Waiting… and render out of order behind a long-running call
- [BUG] - ultrareview failed and consumed free use credit
- [BUG] macOS: npm/Homebrew install places native binary at bin/claude.exe, causing recurring "claude.exe would like to access data from other apps" TCC prompts
- [Feature Request] Restore deprecated Claude Opus versions (4.6, 4.7) with deprecation warnings
- [MODEL] False-positive cyber-safety block on benign Minecraft mod documentation task
- [Feature Request] Background task execution without conversation blocking in VSCode Claude extension
- [Bug] Anthropic API Error: Cannot modify thinking blocks in assistant message
- [FEATURE] Terminal (TUI): inline colored annotations + deferred batch reply to specific blocks of Claude's output (reviving #45904)
- Subagents on Pro 1M tier: trivial probes pass, real workloads fail at first tool call (probe-vs-workload divergence)
- V8 out-of-memory crashes (SIGABRT) freeze/kill the TUI during long sessions on macOS
- iPad browser (vscode.dev): newlines not rendered in message display
- Claude Pro: 1M context limit breaks active workflows without warning
- [MODEL] Opus 4.7 stole my money
- [BUG]
- [Bug] Bash tool executes commands multiple times with duplicate outputs
- [Bug] Tool use regression in version 4.8 causing premature termination and parallel execution issues
- [BUG] Windows: stopShellPty on tab switch kills session with exit code 4294967295 — regression in May 27+ build
- [FEATURE] Multi-LSP per file extension: let linting servers coexist with type servers — or tell us it won't happen
- [BUG] Cowork "Download failed" on macOS — claude-code-vm runtime missing from app bundle
- [MODEL] Opus 4.8 (xhigh): claims work verified when it isn't, self-contradicts, and over-parallelizes contradictory tool calls — on a low-complexity codebase
- [BUG] 400 "thinking/redacted_thinking blocks cannot be modified" bricks session after large multi-tool turns
- Sonnet dispatched via Agent tool fails with "1M context required" error
- iOS Dispatch sessions from mobile fail to spawn reliably
- Dispatch needs user-level persistent persona / identity injection
- [BUG]
- claude can't self test against self bullshit.
- [Bug] Excessive token consumption rate - usage depleting faster than expected
- [BUG] Cowork crashes with exit code 1 on Windows 10 Pro - Claude Desktop 1.18286.0.0
- Desktop app: session list activity dot keeps pulsing on an idle session with no running tasks
- [BUG] Desktop sidebar keeps showing session "running" dot after all background tasks have exited
- [BUG] AskUserQuestion: "No response after 60s — continued without an answer"
- Desktop app: sidebar "Running" indicator stays pulsing on idle sessions (esp. after interrupting turns)
- I need more information to create a GitHub issue title. Could you please provide: 1. What specific problem or error did you encounter? 2. What were you trying to do when it happened? 3. What error message did you receive (if any)? 4. What is the unexpecte
- [BUG] Cowork tab flickers and disappears on Linux beta — yukonSilver not supported (status=unsupported), VM bundle never downloads (AMD Ryzen, kernel 6.17)
- [Feature Request] Deep linking support for MCP server setup in Claude Desktop
- rzoldt- fuck your mother -dwmaoi
- gavmbl- fuck your mother -pwuzgf
- iquunc- fuck your mother -zrzkje
- uysvvc- fuck your mother -xamgzf
- hnpnbx- fuck your mother -ulsklw
- cusbag- fuck your mother -vapwyl
- plctxi- fuck your mother -heuhgo
- ecvzgy- fuck your mother -exride
- flhkzo- fuck your mother -ooquco
- spptsz- fuck your mother -cyofhy
- yobbvn- fuck your mother -uywzpb
- rgldso- fuck your mother -hhrmih
- vfymil- fuck your mother -blqoux
- eaqnga- fuck your mother -xhgibp
- vptdpe- fuck your mother -rlelyv
- yfuhuw- fuck your mother -mmvmua
- gdlcon- fuck your mother -ksrzrm
- mfxugh- fuck your mother -zjskae
- szhdwg- fuck your mother -vcntlc
- euxdvu- fuck your mother -eglfke
- yfriac- fuck your mother -oauoyu
- atqifi- fuck your mother -lopmzs
- whctpo- fuck your mother -zlvsag
- zncwmu- fuck your mother -otmdvz
- ftgjzk- fuck your mother -yarvns
- twqfux- fuck your mother -apvwau
- jxdhfs- fuck your mother -fsmyfk
- lgtzoi- fuck your mother -quxrbh
- fnslfv- fuck your mother -jsxcom
- onrfcb- fuck your mother -aizhal
- gmatmc- fuck your mother -nlxani
- hyqhlc- fuck your mother -lzxwxx
- ejzcqi- fuck your mother -vvpods
- sypssj- fuck your mother -kyfoaf
- opmxik- fuck your mother -irloof
- aerfat- fuck your mother -flnidz
- txbgvy- fuck your mother -atlvss
- qogtyy- fuck your mother -klpyna
- ixuapv- fuck your mother -mmhnbc
- boectb- fuck your mother -lvduyr
- opundp- fuck your mother -lbjerg
- mjwock- fuck your mother -pjmlia
- rcmtzm- fuck your mother -gzufdz
- qmxrkf- fuck your mother -tvybsz
- ybnrzo- fuck your mother -jazkoc
- vngnao- fuck your mother -tthksb
- pdgtfn- fuck your mother -xvhsen
- atkxdz- fuck your mother -apqxpv
- zdmqeh- fuck your mother -wcenwm
- bbimjp- fuck your mother -yyeckb
- yzbhpx- fuck your mother -wipdlo
- oettwx- fuck your mother -yrwqie
- lefxke- fuck your mother -uswmxd
- dtbbcn- fuck your mother -vpojkz
- lchkhg- fuck your mother -zxwwlr
- ieknnh- fuck your mother -ebsefi
- xomzkz- fuck your mother -pspnjj
- iijznb- fuck your mother -eedzlw
- camojm- fuck your mother -dxxurp
- dshhzm- fuck your mother -itimzx
- vojqmz- fuck your mother -lyqict
- twtydj- fuck your mother -pphtut
- hohymz- fuck your mother -xcaxkl
- rocner- fuck your mother -beqzcd
- xiuumt- fuck your mother -hvqdrr
- ppjlph- fuck your mother -zreoqr
- jfyhwt- fuck your mother -iedcjl
- ccmnat- fuck your mother -eawsad
- cvfndf- fuck your mother -tlobts
- vxgkww- fuck your mother -pzuide
- mikooa- fuck your mother -govupv
- rfcqyb- fuck your mother -ingsbi
- jcsswl- fuck your mother -kfxoga
- qybmpm- fuck your mother -nsuqjj
- roeghk- fuck your mother -vcaboa
- Signal page: group parties by entity type (Banks, mREITs, BDCs, Securitizations)
- [Claude Desktop, not CLI] Windows: main process killed at loader time on every launch after first successful run post-install
- [FEATURE] Switch Claude accounts
- [BUG] Cluade failed to resize and image for mobile and almost ruined my entire website
- Foundry: `/context` shows 200k cap even with 1M beta enabled and `effectiveWindow=1M` confirmed
- False positive safety block on legitimate security audit tasks (Supabase RLS review, pnpm audit)
- Feature request: UEStudio (UltraEdit Studio) IDE extension
- [FEATURE] raw terminal output option
- [BUG] feature-dev plugin breaks Ctrl+V (paste) in Claude Code on Windows
- Feature request: j/k vim navigation in interactive selection menus (e.g. /effort)
- [UX] Can't switch model without clearing typed input — need keybinding or inline shortcut
- [Bug] VSCode extension terminal setup writes keybindings to DEFAULT PROFILE instead of selected profile
- False Usage Limit and Request blocked
- Feature Request: Chinese / multilingual UI localization for permission prompts
- [FEATURE] Allow PreToolUse hook output to render before tool execution (non-blocking)
- [BUG] False-positive Usage Policy (AUP) violation on benign FPGA/hardware engineering waveform analysis
- [Bug] Multilingual token mixing in extended context with web search and bash execution
- [Bug] Overly aggressive content filter blocks legitimate infrastructure debugging of timeout issues
- [Bug] Anthropic API Error: Usage Policy false positive issue blocking operations
- False positive on Claude Code Usage Policy classifier — please stop over-flagging legitimate engineering work
- [BUG] False-positive Usage Policy refusal mid-stream on a markdown comparison table
- [Bug] VERY sensitive cyber-safeguards
- [Bug] Usage Policy cyber-safeguards block parallel headless Task() subagents doing localhost QA
- [BUG] System prompt constant qD4 unconditionally injects security-adjacent text into every session, priming the API cyber classifier
- Feature: Hot-reload harness (version/plugins) for running background sessions without losing conversation history
- [BUG] API 500 ERROR "Attempt N/10... Retrying in Xs..." MAKING CLAUDE CODE UNUSEABLE
- [BUG] Cowork omits anthropic-workspace-id header on custom gateway (Code works)
- [FEATURE] Add workflow tool to cloud scheduled tasks
- [BUG] Claude spent 33+ minutes and 18k+ tokens debugging a Gradle build that was actually just an Android Studio version incompatibility
- [FEATURE] Programmatic session rename API for skills/agents
- Project memory bleeds across accounts — personal project context loads in work account session
- [Bug] opus 4.8 hangs indefinitely on tool calls after extended usage, freezes on messages ending with "let me..."
- Claude Desktop window is completely blank
- Remote folder connection fails with "Host denied (verification failed)" for specific hostname
- [DOCS] Per-feature platform availability is not discoverable in docs
- Remote Claude Code session (desktop app): GitHubPrManager spawns local `gh` with remote cwd → misleading ENOENT
- Claude desktop app's PR/CI panel can't find `gh` because PATH is `/usr/bin:/bin:/usr/sbin:/sbin` (LSEnvironment override)
- [Bug] Satisfaction poll UI causes accidental low ratings due to unclear alternatives presentation
- Bash tool silently strips -flag tokens when a command has a command substitution with a nested $
- [BUG] Desktop local-agent (Claude Code): 900s session idle timeout tears down the computer-use connector and never re-registers it on resume
- Cowork Desktop: ${CLAUDE_PLUGIN_DATA} is not persistent across conversations (MCP plugin tokens lost on every new conversation)
- [Bug] Persistent display overlay issue not resolved across multiple prompts
- Usage cap: single ~30k-token prompt (no subagents) burned 38% of 5h Max 20x limit on Opus 4.7
- [Bug] Anthropic API Error: Server temporarily limiting requests due to rate limiting
- [BUG]
- [BUG] Claude hangs with no output on Linux Mint despite all-green
- [FEATURE] Option to disable terminal title override in Cursor/VS Code
- MCP tool calls to api.anthropic.com blocked by Cloudflare WAF (SQL injection rule false positive)
- [BUG] Explorer window opens after installation on Windows
- [BUG] Claude misspells username in working directory path, causing permission prompts
- [BUG] API Error: Unable to connect to API (ECONNRESET)
- [FEATURE] /handover — user-curated session hand-off with fresh context
- [FEATURE] Configurable subscribe_pr_activity wake-message text (settings or hook)
- /model picker does not list newly available models (Opus 4.7, 4.8)
- [BUG] Claude defaults to `.claude` despite having set `CLAUDE_CONFIG_DIR`
- [FEATURE] In-session agent view (← navigation) should be scopeable to the current project
- [Feature Request] Add confirmation prompt before submitting critical feedback
- Illegal Instruction crash on CPUs without AVX since v2.1.113
- dontAsk: Write allow-list broken for all path forms on Windows native (2.1.145); PreToolUse hook confirmed workaround
- [Bug] Desktop Projects "Recent" sort uses stale timestamps instead of chat activity
- [BUG] Sub-agent messages render in wrong vertical order — insertion position resets on main↔sub-agent context switch