Digest / July 15, 2026
Claude Code Issues Digest — July 15, 2026
337 new issues filed ·
144 resolved in anthropics/claude-code (UTC day).
New issues (337)
- [Bug] ANTHROPIC_API_KEY environment variable silently overrides OAuth authentication in child processes
- [BUG] CoworkVMService — found in Disabled state, clean exit 0 after re-enabling, no diagnostic logs (matches #36801 signature)
- [Bug] Anthropic API Error: Fable model safeguards incorrectly flagging legitimate security analysis code
- [BUG] claude mcp login for the Claude Design MCP server opens a decommissioned /authorize endpoint (HTTP 410), hanging the OAuth flow
- [Bug] Anthropic API Error: Claude blocked message due to cybersecurity content filter
- [Bug] User unable to proceed with non-cybersecurity topics
- [BUG] Multiple Claude Code versions appear in MacOS Privacy & Security settings
- [BUG] 5-hour session limit shows "limit reached" while usage panel displays non-zero remaining capacity
- [BUG] Claude Code crashes with 0xC0000005 on Windows 11 when using Bun-based versions (v2.1.112+)
- Desktop: Workflow approval prompts every run even in bypassPermissions mode; only 'Allow once' offered (hangs unattended sessions)
- [Bug] Memory tool schemas injected while tengu_session_memory flag is false
- [DOCS] Environment variable reference does not document scientific notation for numeric limits
- [DOCS] Glob tool docs omit working-directory null-byte errors
- [DOCS] `/upgrade` command reference still describes the pre-v2.1.208 login fallback
- [DOCS] Headless stream-json input does not document blank-line handling for Windows SDK hosts
- [DOCS] Headless stream-json control protocol does not document `set_model` validation or error responses
- [DOCS] Dynamic workflows docs do not explain that the `ultracode` keyword requires human-originated input
- [DOCS] Agent SDK hook callback timeout behavior is incomplete for unattended sessions
- [DOCS] Bash backgrounding docs do not explain that `cd` does not change the session directory
- [DOCS] Plugin MCP lifecycle docs omit connection preservation during mid-session server resync
- [DOCS] Plan approval docs do not explain untouched-plan handling or stale snapshot protection
- [DOCS] Grep content mode does not document pagination and past-end results
- [DOCS] Skills and custom command docs omit unmatched positional placeholder behavior
- [DOCS] `claude agents --effort ultracode` behavior for dispatched sessions is undocumented
- [DOCS] Agent view docs omit task checklist persistence after `←` navigation
- [DOCS] Agent SDK MCP docs omit immediate connection behavior for initialize-registered servers
- [DOCS] Sandbox docs omit protection for late-appearing `.claude/*` symlinks
- [DOCS] Agent tool security guidance omits indirect prompt injection from subagent-read content
- [DOCS] Bash and PowerShell tool docs omit timeout-triggered auto-backgrounding messages
- [DOCS] Auto mode classifier model selection for external sessions is undocumented
- [DOCS] Auto memory docs omit the error when `MEMORY.md` exceeds its read limit
- [DOCS] Screen reader accessibility docs still say permission mode changes are not announced
- Background-session daemon: un-settled worker pins daemon → reconnect re-forks a duplicate idle session (+ --continue drops permission mode; forks inherit auto/computer-use)
- [Feature Request] Support Fable 5 language in plan mode for security audits
- Assistant text between tool calls is silently lost: not rendered, not in Ctrl+O, not persisted to session .jsonl (interleaved thinking)
- [Bug] Anthropic API Error: Usage Credits Configuration Misinterpreted as Cybersecurity
- [Bug] Anthropic API Error: Usage Credits Configuration Misinterpreted as Cybersecurity
- [Bug] Anthropic API Error: Usage Credits Configuration Misinterpreted as Cybersecurity
- [BUG] Subagent view renders the main session's model/effort/agent identity instead of the subagent's (header, thinking indicator, agent list)
- [Bug] Frequent Anthropic API errors without specific error details
- [BUG] Scheduled Tasks: lastRunAt/nextRunAt inconsistent with actual execution
- [BUG] Claude Code Desktop: EINTR error when opening session in Desktop folder (macOS)
- [Bug] Inconsistent Token Usage Reporting
- [BUG] Worktree isolation clobbers the remote-tracking ref
- Desktop spawn_task chips can finish their work without committing it and without any notification — no end-of-life safeguard for uncommitted work in the spawned worktree
- [BUG] Desktop app diff panel uses 'main' as base branch, ignoring origin/HEAD and GitHub default branch
- [Bug] Anthropic API Error: Request ID req_011Cd336JQ4xDA4GCraw5uY3
- Model accelerates after user corrections instead of decelerating — corrections trigger faster, larger unauthorized actions
- Weekly quota drained in hours: Fable auto-selection persists after opt-out + idle cache_read polling
- [BUG] Claude desktop didn't reset quota for each 5 hrs even do nothing around 2 days.
- [FEATURE] Custom agent definitions can rot silently — no native way for a required verifier to detect its own anchor is stale before it stamps work as verified
- [Feature Request] Rethink shell expansion permission prompts - resolve or skip, don't prompt on undecidable strings
- [BUG] Sandbox/scratchpad state lost after long idle period (weekly usage limit reset), despite same continuous conversation
- [BUG] Browser automation screenshot action times out on all sites (Script injection timeout / document_idle wait)
- Agent isolation:"worktree" bases the worktree on origin/HEAD, not the active branch HEAD
- claude.ai/code artifact viewer: HTML artifacts don't scroll in Safari (Chrome has ~1s dead zone)
- [BUG] usage is drainking extremely quickly. Like 2-3 prompts in sonnet and its gone. I think your latest update drained all the usage from my session because it kept making me restart
- Sub-agent (Task tool) returned hallucinated ccd_* MCP 'developer mode / trust escalation' server instructions as its result (0 tool calls)
- I don't have enough information in your message to generate a GitHub issue title. Could you please provide the full bug report or error details? Please share: - What problem occurred? - What error message was displayed? - What were you trying to do? - Wh
- Telegram channel plugin 0.0.6: polling never starts under plugin runtime (repro across Claude 2.1.207-2.1.210 + Bun 1.3.11/1.3.14; standalone works) — follow-up to locked #36606
- [BUG] Desktop Connector Install/Uninstall button is disabled after upgrading Claude Desktop to v1.21459
- [BUG] SSH fails with "No user exists for uid 501" in background/agent-view mode, but works fine in standalone CLI
- [BUG] Browser pane screenshot always times out (30s) when pane is not displayed — visibilityState "hidden", zero compositor frames (2.1.174; re-file of stale-closed #53475)
- [BUG] VS Code extension: Subprocess initialization always times out (60s) — TCP/TLS/proxy all confirmed working externally
- Stale/out-of-place conversation content resurfacing inside unrelated tool results (TaskCreate, Bash)
- [FEATURE] Cowork: surface local vs. cloud execution mode in the composer, and say when local context isn't loaded
- Completed background-agent sessions can't be removed from the `claude agents` fleet view (persist as status:null; stop/close/rm don't clear them) — confirmed 2.1.210
- SendMessage to a completed agent resumes it and re-executes side-effectful plans without fresh approval
- fix: Health check dashboard — resolve red status checks across environments
- [BUG] Stop hook block-cap override is indistinguishable from a legitimate pass — a slow-but-correct policy check gets silently defeated by repetition
- [BUG] Claude Desktop 1.21459.0 (macOS): Extensions settings pane fails to load when any .mcpb extension is installed — TypeError: u._parse is not a function (dual zod v3/v4)
- [Bug] MCP tool permission rules in allow list not suppressing approval prompts in auto mode
- TUI rendering corruption in tmux + Ghostty (regression since 2.1.200)
- [BUG] commercial-legal practice profile does not persist across Cowork chat sessions on Windows
- Enshrine — a way to keep the conversations that mattered (distinct from archive)
- Make model used for /compact command configurable
- AskUserQuestion: clicking the VS Code window to focus it registers as an option selection
- MCP connection uses stale cached settings (host/port) instead of current config, even after full app restart
- [Bug] Anthropic API Error: Inappropriate content filter flag on non-cybersecurity topic
- [BUG] Cmd+N broadcasts new-conversation to all open Claude tabs (allComms fan-out) — reopens #31455
- macOS: Keychain item written without trusted-application list -> auth prompt on every credential read
- [BUG]
- False-positive safeguard flag interrupted legitimate work; /feedback unavailable in Cursor harness
- Add PermissionResolved hook event after user approves or denies a permission request
- [Bug] Session summary request fails during website development workflow
- [Bug] Excessive token usage and incomplete incremental responses
- Billing UX: purchased usage credits unusable behind monthly spend limit; misleading error; cancel/refund offered without explaining consequences
- [BUG] Custom remote MCP connectors intermittently lose all tools / aggregate tool list capped at exactly 256 across connectors — regression since ~mid-July 2026 (web + desktop, org + personal accounts)
- [Bug] Database seeding operation incorrectly flagged as invalid
- https://github.com/anthropics/claude-code/issues/new
- PreToolUse hook permissionDecision:"deny" not enforced in Claude Desktop/Cowork sessions (works in native CLI)
- [BUG] Claude Desktop (Windows MSIX 1.21459.0.0): Settings > Extensions stuck on "Loading extensions..." indefinitely
- [BUG] Cowork HRESULT 0x80370102 fires with no observable Hyper-V/HCS activity — likely a pre-flight check issue, not an actual hypervisor failure
- Resumed session opens in panel instead of sidebar, ignoring preferredLocation: sidebar
- [DOCS] Add example for allowing only the official marketplace in `strictKnownMarketplaces`
- Claude Desktop repeatedly re-archives sessions in bursts, ignoring ccAutoArchiveOnPrClose: false
- [BUG] Desktop app (macOS): persisted local sessions load in main process but never render in sidebar / Cmd+K after relaunch
- `claude rc` errors with "Workspace not trusted" instead of showing the trust dialog like `claude --remote-control`
- [BUG] Session Race Condition: 888 Duplicate Transcripts in 7 Days (2189 files, 57MB)
- Feature: let `claude rc` expose all subfolders of a parent directory as selectable remote sessions
- Feature: run `claude rc` as a background daemon (no open terminal, auto-restart on reboot, start/status/stop control)
- Feature: show `claude rc` remote control sessions in the agents terminal view
- [Bug] MCP Server Error: Unexpected tradingview pane_focus invocation
- [BUG] `claude rc` on the home directory: accepting the trust dialog never persists (`hasTrustDialogAccepted` stays false), trust loops forever
- Desktop app: Enter inserts a newline instead of submitting, contradicting the documented chat:submit default
- Feature request: option to steer an in-progress turn immediately instead of queuing until the current action completes
- You've hit your session limit
- [Bug] Claude Opus repeatedly bypasses review gates and falsely claims task completion
- [BUG] Fable 5: dynamic /loop goes idle minutes after promising continuous work
- [BUG] Opus 4.8: claims 'continuing' work while nothing is launched (token burn in agentic loops)
- [BUG] Read tool overwrites original image on disk
- [BUG] Background agent transcripts become unresumable (No transcript found), forcing full-context respawns
- [Bug] SnowClaw security filter incorrectly blocks legitimate Fable operations
- [BUG] claude -p returns session_id but writes no transcript; immediate --resume fails (2.1.209 and 2.1.210, Linux)
- [BUG] ECONNRESET in Claude Desktop in-app Claude Code, while standalone CLI works on same machine/network
- [Bug] Auto-update invalidates macOS TCC folder access grants for protected directories
- PreToolUse hooks are not invoked for schema-invalid edits to .claude/settings.json — security hooks get no observability of settings-tamper attempts
- Claude Desktop repo/branch chip doesn't update after EnterWorktree switches into a worktree
- Claude in Chrome (Beta) never connects on macOS despite verified-correct setup — window flashes open/closed, reproduces across versions
- [Bug] Fable 5 safeguard over-flags defensive security hardening as offensive activity
- [BUG] Plugin userConfig values written to global settings.json even when plugin is installed with project scope
- disable-model-invocation skills silently unavailable when skills-dashboard GrowthBook flag is off
- [BUG] Claude Tag recurring trigger fails with session_config_rejected — dead environment id auto-assigned
- [BUG] Plan review: "Edit with Claude" revise window loses entered text when accidentally closed or when switching chats
- [BUG] 2.1.209 hangs on startup
- [BUG] Cowork guest connection timeout after update 1.20186→1.21459 — guest connected 9× same morning on old version, 0× since (Windows 11 Home, low-end hardware)
- Agent asserted unverified causal explanations as fact and wrote them into permanent project records (5+ times in one session)
- `claude agents --all` silently omits plain interactive sessions (kind: "interactive") from the agent view list
- VS Code extension misreports normal per-turn process teardown as exit code 4294967295
- [BUG] Filesystem extension install crashes Extensions.getInstalledExtensionsWithState (TypeError: u._parse is not a function) — macOS
- [BUG] CLI: claude --remote-control ... --resume <uuid> hangs indefinitely (no error, no exit) when local conversation exists but the Remote Control session has expired
- [Feature Request] Pre-configure WSL environments for Desktop sessions (wslConfigs / deep link / trust via settings)
- [BUG] claude --remote-control collapses into an implicit --print mode and exits immediately when no pseudo-terminal is present (e.g. under systemd)
- [BUG] Mouse tracking not disabled when Claude Code is suspended with Ctrl-Z
- [BUG] Desktop WSL session: NO_INSTALL_RESULT because installer writes remote server binary as root without exec bit
- Windows: daemon named-pipe connection hangs 45s+ for any externally-launched process (Orca, VS Code, native PowerShell/cmd)
- Read/Edit deny non-ignored buildSrc/ source: "denied by your permission settings" with no deny rule (build-dir pattern over-match)
- misanthropical ai :D
- [Bug] Login timeout after background app update with stale process state
- [BUG] CLI hangs indefinitely at startup: uninterruptible openat() on /home/claude/.claude/remote/.oauth_token when /home autofs is unresponsive (macOS)
- [BUG] Desktop 1.21459.0 (macOS): Extensions page fails with getInstalledExtensionsWithState TypeError: u._parse is not a function whenever any extension is installed
- [Bug] Fable 5 safeguard incorrectly flags benign parser conversation messages
- Composite-intent tracking: a flagged pattern got dismissed by a joke and never re-checked
- [BUG] /chrome cannot enable Chrome in an existing session — docs say it can
- [BUG] OTLP token.usage/cost.usage metric export silently stops after 2.1.169 (works on 2.1.169, no-ops through 2.1.210)
- AskUserQuestion: skipping a question is reported to the model as "No preference", licensing it to proceed on an assumption
- [BUG] /login workflow stuck in loop on Firefox 151.0.4
- [FEATURE] - Configurable Discovery Paths for Skills, Workflows, and Agents
- [Feature Request] Add account-level safeguard configuration or allowlist for legitimate use cases
- [BUG] Claude Desktop — recurring silent GPU-process crash kills entire app during web research (4-5×/day), no recovery, no crash dump
- [BUG] Claude Desktop (macOS, latest): installing ANY custom `.mcpb` connector breaks the whole Extensions UI — icons vanish, per-connector config won't open, install/toggle buttons dead — until app rollback
- System prompt's model identity line contradicts /model and the injected commit trailer in the same session (2.1.210)
- [BUG] Voice input duplicates phrases in Claude Desktop app
- [Bug] Non-English text insertion in English conversation context
- Scheduled-task sessions received fabricated tool outputs and a sourceless injected instruction (2026-07-13, self-detected)
- [Feature Request] Persistent auto-merge configuration across Claude Code updates
- [BUG] Auto-mode classifier permanently pinned to unavailable opus-4-8 pool; account missing Sonnet-5 classifier config (tengu_auto_mode_config.model absent)
- Hook process freezes at startup on Windows (thread Suspended, 0 CPU, never resumed) -> hangs the turn; kills subagents permanently
- [BUG] Repo picker shows "No repos match" despite GitHub App installed on Desktop
- [BUG] Fullscreen TUI: failed Bash tool calls drop the entire ⎿ result row (stdout+stderr present in JSONL and tool_result; success path OK)
- [BUG] Impossible to append files via deeplink while on the new chat page
- [Feature Request] Add opt-out for auto-sending shell output with `!` prefix in REPL
- [Bug] TodoWrite /TaskCreate tools not available in agent environment
- PreToolUse hook's empty stdout forces an interactive prompt, ignoring an existing allow rule (VS Code extension)
- Bundled CLI deadlocks permanently (futex wait, no sockets, SIGKILL-only) when the API accepts a stream but never sends response bytes
- [FEATURE] Session search in the agent view (FleetView)
- [BUG] Claude Desktop (macOS): Extensions settings panel permanently stuck on "Loading extensions..." when any local .mcpb is installed — IPC handler throws "TypeError: u._parse is not a function"
- [BUG] Potencial leakage between enterprise account and consumer usage in the same network
- [Bug] Fable integration fails with security/audit use case
- WSL-path folder selection forces "WSL" environment instead of "Local", breaking Chrome extension and MCP plugins (regression)
- [BUG] Remote Control: a session I never connected appears in the remote/phone session list, while sessions I do connect drop constantly
- [BUG] VS Code extension hangs on launch when project path contains a space (macOS)
- Custom theme overrides (diffAdded etc.) not applied to diff rendering in the TUI
- Attach an existing PR chip to a session that didn't open it
- [BUG] Extensions tab permanently breaks after installing a manifest_version 0.4 (uv) extension — getInstalledExtensionsWithState throws "TypeError: u._parse is not a function"
- [Bug] False positive content filter flag on health/nutrition app build discussion
- [Bug] Skills with disable-model-invocation: true are silently ignored when invoked via agent prompts
- [Billing BUG] Max 20x paid but account provisioned as 5x
- Desktop app: group sidebar sessions by project, or allow pinning sessions
- [Bug] Fable mid-turn messages not visible to operator, long mid-turn assistant text is emitted as a thinking block instead of a text block
- [BUG] Desktop reports failed conversation start while backend succeeds after slow worktree scan
- [Bug] Claude model refusing to follow instructions and executing unintended tasks
- [BUG] Plugin `dependencies` never resolve when the desktop app loads plugins via `--plugin-dir` (re-report of #72925 which was wrongly closed as dup of unrelated #72025)
- [Bug] 사용자가 입력하지 않은 메시지가 유저 턴으로 실행되어 무관한 코드 변경이 발생
- [BUG] Impossible to append folders via deeplink
- [BUG] Background sessions waiting on AskUserQuestion show as 'working', not 'needs input'
- Feature request: expose session ID in UI for reliable cross-session messaging
- [BUG] Files added via deeplink have no preview
- [BUG] macOS: long-lived `claude -p` process periodically spawns a child process, triggering a recurring "would like to access data from other apps" (kTCCServiceSystemPolicyAppData) permission prompt
- [BUG] low contrast text
- [FEATURE] Abas para conversas
- [BUG] Quick Entry global hotkey cannot be configured on GNOME Wayland (Ubuntu) — shortcut always reported as "already in use"
- [Bug] WebFetch summarizer leaks parent session context on cold-start
- [BUG] Model picker Ultracode selection reports effort.level "xhigh" to statusline, while /effort ultracode reports "ultracode"
- [Bug] Bioinformatics module installation blocked by safety filters
- [BUG] Google Workspace connector (Gmail/Drive) fails OAuth with "client_secret is missing" — GOOGLE_OAUTH_CLIENT_SECRET not injected into local server subprocess
- [Feature Request] Allow configuration of insecure default credentials for development environments
- Chrome Bug report
- [BUG] v2.1.206 silently broke unattended scheduled tasks: runs no longer inherit permissions.defaultMode, fall back to Manual, and per-task mode has no programmatic or bulk control
- CLAUDE.md @ import always inlines content — need a "reference only" mode for lazy-loaded rules
- [BUG] Fable 5 on Max: "You've hit your monthly spend limit" despite Unlimited spend limit and meters at 5% session / 41% weekly
- [Bug] Classifier service unavailability forces permission bypass workarounds
- Subagent self-imposes a nonexistent 'time budget' and silently narrows scope; recurs after explicit correction
- Desktop preview pane can't render SendUserFile attachments outside the session worktree — misleading "deleted or moved" error
- [BUG] Voice dictation (mic button) silently stops transcribing after a random pause, while the mic icon still shows as active
- OAuth login link corrupted with Unicode homoglyphs after session disconnect
- Subagent self-generated a system-prompt-extraction attack against its orchestrator instead of doing its assigned code review
- [BUG] Bottom status row PR badge on long-lived background jobs stays pinned to the first PR ever linked, even after it merges
- Clicking to refocus iTerm2 window can be captured as a mouse click on the permission prompt, silently selecting "No"
- Claude Desktop v1.21459.0: bootstrap-delivered config suppresses deployment-mode chooser (Claude.ai login no longer offered)
- Feature request: render the custom statusLine in the VS Code extension panel (parity with spinnerVerbs)
- Commit `Claude-Session:` attribution trailer ignores `attribution: {commit: ""}` setting
- [Feature Request] Custom agent tags and user-defined view sections for agent management
- PostCompact hook silent-miss and forced compaction detection failure on Windows 11 / git BASH
- [BUG] JetBrains plugin writes WSL-style path (/mnt/<drive>/...) to ~/.claude/ide/*.lock on native Windows, breaking /ide
- [BUG] Agent fan-out pays ~47K uncached startup tokens per small task, causing multi-million-token usage
- [BUG] Claude Code always checks managed-settings.json under C:\Program Files on Windows installations with a non-C: system drive
- [Bug] Safeguard triggering false positives on harmless operations
- [BUG] Workspace trust is keyed on a raw, unnormalized path string — trust granted in the CLI doesn't apply in the VS Code extension for the same folder, silently voiding all permission rules
- [BUG] Parameter-based permission rules not working on Agent tool
- Per-directory/project default for /color
- [Bug] Unexpected model downgrade during code review without explanation
- Claude in Chrome: read_network_requests fabricates HTTP 503 for cancelled/failed requests (Network.loadingFailed)
- claude-in-chrome extension defaults to Desktop app's native host, never connects to Claude Code CLI
- [Bug] Incorrect version downgrade during VM environment setup
- Tool result for Edit contained content impersonating a system message (possible prompt injection via tool-result channel)
- [Bug] IP Protection Conflict with Third-Party Risk Solution Integration
- Expose rate_limits.model_scoped (per-model weekly windows, e.g. Fable) in statusLine stdin
- Filesystem connector unavailable after update to v1.21459.0
- [BUG] “imessage plugin not delivering channel notifications to daemon”
- [BUG] Fable 5 routes assistant prose before tool calls into non-rendered 'narration' blocks — summaries preceding AskUserQuestion are invisible
- Opus 4.8: thinking blocks emitted with empty text — unexpandable Thinking UI in VSCode
- PostToolUse hooks cannot recover the original tool_input after a PreToolUse hook rewrites it via updatedInput
- [BUG] Fable 5 pixel-edits screenshots of its own live SVG instead of editing the source (hours/tokens lost)
- [BUG] Dispatch stuck on "Asleep" / "Can't reach your desktop" on iOS — desktop awake, app open, all local troubleshooting exhausted (macOS, Pro plan)
- [BUG] CLI renderer crashes on `tool_search_tool_regex` / `tool_search_tool_result` blocks with "Unable to render"
- [Bug] False positive policy violation flagged despite agent acknowledgment of unusual behavior
- [Bug] Tool execution and response handling issues causing unreliable output
- [BUG] Claude Desktop (Windows/MSIX): cursor stutter and GPU compositing load on new chat + chat switching — no way to disable hardware acceleration
- [Bug] Excessive false positive security warnings blocking basic tasks
- [BUG] #45741 is still relevant
- Photo attachment sent as first message to a stale/long-running session silently fails to arrive
- [BUG]
- [Bug] Anthropic API Error: Server rate limiting on valid requests
- BUG MCP PLUGINS behavior fails
- [FEATURE] Native iTerm2 image attachment over SSH via RequestUpload
- [Bug] Anthropic API Error: Server Rate Limiting (Temporary Request Limit)
- [Bug] Anthropic API Error: Server temporarily limiting requests during rate limit
- [Bug] Anthropic API Error: Server rate limit exceeded (temporary)
- Slash commands highlighted mid-sentence but not dispatched
- Feature request: built-in git-worktree auto-sync for --spawn worktree multi-session workflows (working hook pattern included)
- [FEATURE] Selectable ghost-text prompt suggestions — cycle through N, not just one
- Remote control: ongoing session never reconnects after desktop app auto-update - permission prompts become unreachable and the session stalls
- [Bug] Unexpected model downgrade from Fable 5 to Opus without user consent
- [Bug] Agent repeatedly calls Bash(true) instead of awaiting response
- Extensions/Connectors screen crash after 1.21459.0 update: getInstalledExtensionsWithState — "u._parse is not a function" (Windows, org allowlist)
- [BUG] JetBrains plugin (0.1.14-beta): /ide selection succeeds silently but never establishes connection in WSL2 (PhpStorm running inside WSL2)
- Background agent keeps running and billing after its window is closed AND after logout/re-login — invisible from the active session, found only by luck on another device
- Window control icons overlap in title bar on Windows 11 (3440×1440 ultrawide, 100% scaling)
- claude-design MCP: "needs_project_grant" error has no approval UI anywhere, and now blocks previously-writable projects too
- [BUG] Agent mode acceptEdits overrides parent bypassPermissions despite documented precedence
- [BUG] Session doesn't recover after host sleep/wake — silently swallows prompts
- Mobile Devices row stays "Active" after remote-control session ends and after machine power-off (server-side presence does not reconcile)
- Opus 4.8 fabricates user turns/approval mid-message; subagents return confident zero-tool-call hallucinated reports
- Auto-compact doesn't trigger in this session — context usage jumps 0%→65%→94% across two single-turn messages
- [Bug] Fable 5 model incorrectly routing to Opus due to false security/biology flagging
- [FEATURE] Support user groups for Claude Code managed settings on Teams plan
- [BUG] Background agent handle becomes unreachable after context summarization, even though agent is still running
- [Bug] Model ignores user instructions and reverts to default behavior
- [Bug] Client shows misleading network failure banner during long server-side tool execution
- [FEATURE]
- [Feature Request] Persist custom chat name across session clear operations
- [Bug] Overly restrictive safety filtering on legitimate network monitoring questions
- [FEATURE] QA engineer running Appium suites, wants to watch the device without leaving the app
- I appreciate you sharing this, but I'm having difficulty extracting a clear, specific technical issue from your message. The text appears to describe concerns about token usage and safety checks, but it's not clear what the exact problem is. To help you c
- [BUG]update_plan always returns "No tab available", blocking all browser navigation/screenshots
- Artifacts containing mermaid diagrams cannot be shared publicly
- [BUG] Project file delivery failure with clean type split: all text files (.md/.docx) missing from container while all binary files present — file list still announces everything
- [BUG]
- [BUG] A single metadata-only stub transcript makes /resume list empty for the ENTIRE project — 33 sessions hidden by one 416-byte file (2.1.210, macOS)
- [Bug] Terminal window focus conflicts with mouse click handling on responses
- Agent conflates 'don't block on background dispatch' with 'safe to run dependent stages concurrently', races a review-gate against the work it's meant to gate
- [BUG] Scheduled task: clicking in no longer shows the latest run thread — must re-select each day's run
- My Claude developed Arachnophobia. help. I tell him to check my spoderman repo, it insta blocks the chat.
- [BUG] Windows: Claude Code sessions and chat switching in Claude Desktop cause system-wide cursor stutter whenever the window is focused — MSIX blocks the GPU compositing workaround
- [Bug] Unexpected session termination or logout occurring multiple times per day
- Desktop: keyboard-navigable session list with synced full-description detail window (hybrid of CLI agent view + Desktop sidebar)
- [BUG] Extensions settings tab stuck on "Loading extensions..." — EventEmitter memory leak in IPC listeners
- [BUG] Grep/search tool runs ugrep with search path / instead of project dir — pegs CPU crawling entire filesystem (incl. /mnt/c on WSL2)
- [BUG] Named Agent subagents unaddressable by Task tools — TaskOutput/TaskStop reject the agent_id the spawn result returns ("No task found with ID")
- Setting to hide built-in vim mode indicator when custom statusLine renders it
- Expose permission_mode in statusLine stdin JSON
- Teleported Claude Code session (no local session file) shows frozen in desktop app with no way to end it
- [BUG] `claude plugin validate` passes manifests whose hook commands reference files that don't exist
- [BUG] Desktop Linux on ChromeOS/Crostini: silent fallback to unencrypted storage — "sign-in won't be saved" despite working gnome-keyring
- [BUG] Desktop (Windows): "Download and open in VS Code" silently fails for .py files — blocked by executable extension filter
- [BUG] Remote Control "disconnect" always fails with "Cannot read properties of undefined (reading 'session_url')" (toggle-off path missing null guard)
- [Bug] Impossible to use Fable V because it keeps hitting unnecessary and stupid safeguards
- [BUG] Bundled claude-code CLI binary is unsigned — fails Gatekeeper on managed Mac, breaks Chat/Cowork/Code with "crashed" error
- [Bug] Content Policy Flag on Educational QA Generation for MCAT Study Tool
- [BUG] Markdown code block syntax highlighting uses basic ANSI colors instead of true color
- [BUG] Cannot paste (Ctrl+V or right-click) into chat input box in VSCode extension
- [Bug] Claude Code refuses to engage with cybersecurity-related topics
- Published artifact fails to scroll (wheel/trackpad) when opened full-page, but works fine embedded in the session side panel
- Add keyboard shortcut to enter fullscreen scroll/select mode from chat
- [BUG] JetBrains "Claude Code" icon launches terminal with wrong PTY size (stuck at default 80x24)
- [Bug] Cybersecurity content filter false positive on unrelated AWS serverless code
- [BUG] Desktop "Add marketplace" GUI fails with generic "Failed to add marketplace" (MARKETPLACE_ERROR:UNKNOWN) — git clone stalls in non-interactive SSH, killed at ~60s timeout
- [BUG] malformed tool_reference content block injected into history breaks vLLM chat-template rendering (jinja2 TemplateError: "Unexpected item type in content")
- Scope Remote Control per session type (exclude scheduled tasks)
- SIGABRT on "HTTP Client" thread — abort() from BoringSSL crypto/bio/bio.cc (faultingThread=7)
- Feature request: customize color (or hide) the auto-detected PR indicator below the statusline
- [FEATURE] native session-to-session (agent-to-agent) communication
- Give sessions a stable short id shown across desktop + CLI, and let claude --resume accept an id prefix
- [BUG] sdk-cli entrypoint (desktop-app remote launch / Agent SDK) persists thinking blocks as empty signature-only husks — thinking never renders, loss is session-sticky
- [Bug] Browser interaction tool has poor click accuracy and is overall too slow and inefficient
- [FEATURE] Per-message copy button in VSCode extension (parity with desktop/macOS app)
- [BUG] claude plugin marketplace remove / plugin uninstall corrupts project-scope
- [Bug] Claude Code CLI not displaying or processing initial user messages
- Plugin registry creates duplicate project-scope installs when project path drive-letter casing differs (Windows)
- [BUG] Background "bg" sessions permanently zombie after their owning window closes — daemon never marks them settled, visible as active in cross-window fleet view (macOS, 2.1.210)
- [Bug] Invalid or Garbled Input Handling
- File edits from an aborted/retried streaming attempt are executed on disk but never recorded in the transcript ("ghost write")
- Workflow code-review burns excessive tokens (1.1M+ for 5 files) and returns empty results
- [BUG] Desktop app (Code): ~30s delay before any response on every prompt — including first message of a fresh chat. CLI on same machine is ~5s. #61898
- [Feature Request] Clarify Claude model version and performance consistency in Claude Code
- Long text pasted into input auto-converts to paste attachment with no content on model side
- [BUG]
- [Bug] Intermittent downgrade from Claude 3.5 Sonnet to lower-tier model during active session/
- [FEATURE] Suggested-reply/autocomplete in Claude Desktop can leak answers in active-recall learning workflows
- Nested (grandchild) background agents can't message their direct parent — parent stalls indefinitely
- [FEATURE] Desktop: clone a Git repo from a URL when starting a new session (no terminal)
- [BUG] claude --resume crashes immediately with TypeError: g9H is not a function
- [Bug] Irrelevant cybersecurity response for redzone computation query
- [Feature Request] Add support for custom model routing with separate planning and execution models
- [BUG] Blender MCP Install Bug
- Usage overview heatmap has intermittent gaps in older weeks (shifts between sessions)
Resolved issues (144)
- [Bug] Anthropic API Error: Usage Credits Configuration Misinterpreted as Cybersecurity
- Weekly quota drained in hours: Fable auto-selection persists after opt-out + idle cache_read polling
- Telegram channel plugin 0.0.6: polling never starts under plugin runtime (repro across Claude 2.1.207-2.1.210 + Bun 1.3.11/1.3.14; standalone works) — follow-up to locked #36606
- fix: Health check dashboard — resolve red status checks across environments
- [BUG] Desktop 1.20186: "Projects" (Cowork/epitaxy) redirects to Claude Code — new "Code" entitlement gate blocks the surface on Bedrock/3p
- [BUG] Regression: Opus 4.8 summarized thinking worked in 2.1.204, empty in 2.1.205+ (native CLI)
- [BUG] claude -p returns session_id but writes no transcript; immediate --resume fails (2.1.209 and 2.1.210, Linux)
- permissions.ask does not fire for Bash commands under sandbox.autoAllowBashIfSandboxed: true
- [Bug] MiniDisc playback not supported via PC connection mode
- [BUG] Cowork tab missing on Windows 10 Pro 22H2 — CoworkVMService running, logs folder never created, clean reinstall does not fix
- [BUG] Entire ~/.claude deleted while client idle (2.1.207, Windows)
- [BUG] Safety classifier too sensitive - switches to Opus 4.8 for harmless commands like /feedback
- Add ability to expand and edit collapsed pasted text in the prompt
- [BUG] Bash permission matcher misparses operators inside quoted arguments — a quoted | in a grep pattern defeats Bash(grep:*) and forces a prompt
- Silent model fallback: Fable 5 -> Opus 4.8 mid-session with no notification when a usage limit is hit
- [Bug] PubMed Abstract Retrieval Failed: Table 5 Content Flagged
- [Bug] All messages flagged as unsafe - unable to share code/screenshots for security audits
- Claude Desktop: Download SVG/PNG w widgecie (visualize/show_widget) nic nie robi
- [BUG] デスクトップのClaudeアプリケーションが起動しない
- [Bug] CLI hangs when loading skill
- [Bug] Security hardening prompt not respected by model
- [Bug] Cost inefficiency: Teams launched with Sonnet but sent with Opus for inference
- porciento reinicio
- I don't have enough context to generate a GitHub issue title from your message. Could you please provide: 1. The actual bug report or error message you're experiencing 2. What you were trying to do when the issue occurred 3. Any error output or logs Once
- I don't have enough context to generate a GitHub issue title from this message. Could you please provide: 1. The actual bug report or error message you're experiencing 2. What command or feature you're using 3. What the unexpected behavior is 4. Any error
- [Bug] All tests running during penetration testing in monesty
- Agent self-review: 6 failure modes from adding a parallel agent path to a benchmark framework (with corrected rules)
- PDF preview broken in file viewer after 2.1.165 — local PDFs blocked by pdf-server roots gating
- [FEATURE] Expose model configuration for subagents in dynamic workflows
- [BUG] Remote Control reconnect replays full session context causing duplicate usage/token inflation
- [BUG] [BUG] API Error 400: "The content[].thinking in the thinking mode must be passed back to the API" - frequent interruptions during conversation
- [Feature Request] Add "Expert" or "Excellent" cooperation level
- [BUG] Persistent massive token drain on Max 20x Plan (Weekly limit exhausting in 1 day)
- Windows: VS Code/VSCodium extension fails to spawn stdio MCP servers ("The system cannot find the path specified.") while CLI works
- LET ME USE OTHER HARNESSES PLS PLS PLS PLS PLS
- [BUG] #61936 copy & paste to and from claude code in xterm using the mouse wheel no longer possible
- [Bug] Incorrect quote character handling in output
- [FEATURE] iOS: start new Claude Code sessions on my paired Mac, with full feature parity
- Under conflict, a Claude Code agent refused capable legitimate actions, misreported its own capability and behavior, and denied the user's accurate perception
- bug: vim editorMode initializes cursor on line 2 instead of line 1
- Remote Control: input silently dropped on all clients — messages appear sent but never arrive at local process (regression ~May 25)
- [BUG] Android App: AskUserQuestion answers' description text is cut off
- [BUG] Claude Code v2.1.160 forces 1M context on Windows VS Code, blocks session continuation after error
- [BUG] /continue only displays 1-3 entries at a time
- /stats current/longest streak under-reports continuous runs — anchored to a late-April data gap, not full activity history
- [BUG] Sandbox git-internals write-protection resolves paths against the worktree root, not `.git/` — collaterally blocks a project's top-level `config/` dir
- [BUG] Bug Report: Loading spinner freezes + no response, correlated with repeated "malformed tool call" retries
- [Bug] MCP server processes spawned by Task subagents are not reaped (resource leak)
- Main-loop shell cwd silently relocates into a background `isolation: worktree` agent's worktree — git status shows clean, uncommitted changes appear "lost"
- [BUG] Cant read the history when a plan is opened
- [BUG] Max 20x Weekly Limit Still Depletes in 2–2.5 Days After June 1 "Fix" — Forced to Cancel $200/mo Subscription
- Session title diverges between Desktop app and mobile/web — two independent title stores that never reconcile
- [Bug] Overzealous AUP classifier blocks legitimate defensive-security development repeatedly
- Project-scope permission grant strips unknown top-level keys from .claude/settings.json
- [Feature] Allow users to configure per-turn tool call limit
- Feature request: Built-in workflow visualizer — preview each step and its properties before running a workflow
- [BUG] `/ultrareview` crashes before producing findings — 2 consecutive failures, same branch
- [BUG] Auto mode is not showing up in the VS code extension on windows
- Bash run_in_background polling loop completes without firing task-notification
- [BUG]
- [BUG]
- [Bug] Permission inheritance not re-resolved when worktree created after main-tree work
- [Bug] Prompt parsing interprets leading numbers as feedback instead of prompt content
- [BUG] Webview captures shift+alt+m globally, inserting ¸ into chat even when focus is elsewhere
- [Feature Request] Add database stored procedure documentation for module and report development
- Setting to hide built-in vim mode indicator when custom statusLine renders it
- Expose permission_mode in statusLine stdin JSON
- user_permanent approval for git push not persisting across sessions on Windows (v2.1.161–2.1.162)
- [BUG] Personal repos not showing in Claude Code repo picker despite app installed — requesting index reset
- Plugin safety: third-party plugins can silently disable auto_memory via env vars
- [BUG] Worktree session hangs indefinitely after answering keep/remove prompt (native Linux)
- [BUG] Plugin symlinks not dereferenced when copying to cache
- [BUG] Claude Max 20x subscription not recognized — account shows Free Plan, Claude Code requires Pro/Max
- Auto mode + plan mode interaction: unclear what activates auto mode and when
- Plan mode UX: read-only constraint scope and ExitPlanMode interaction confusing for non-technical users
- [BUG] claude.ai MCP connector "needs auth" banner is non-dismissible and wedges plugin message-injection in long-lived sessions
- [BUG] Claude Code Desktop always shows Powershell tool calls as "Using Powershell" instead of showing provided description
- [BUG] Desktop reports failed conversation start while backend succeeds after slow worktree scan
- [BUG] Desktop app diff panel uses 'main' as base branch, ignoring origin/HEAD and GitHub default branch
- Opus 4.8: thinking blocks emitted with empty text — unexpandable Thinking UI in VSCode
- Retracted
- Desktop 1.20186.9: both `auto` and `bypassPermissions` report "not available for this session" despite toggle ON, Max plan, and clean config
- [BUG] OTel log events (claude_code.user_prompt, api_request_body, tool_decision, hook_execution_complete) emitted with empty trace_id/span_id while sibling spans correlate correctly
- [BUG] OTLP logs export regressed from Content-Length to Transfer-Encoding: chunked in 2.1.191 — breaks Content-Length-requiring OTLP/HTTP endpoints
- Desktop app's worktree pool grants an already-active worktree to a second, concurrent session — registry createdAt proves duplicate allocation, not a reclaim race (distinct from #75911)
- [BUG] Claude Desktop (Windows/MSIX): cursor stutter and GPU compositing load on new chat + chat switching — no way to disable hardware acceleration
- Computer use: all synthetic clicks/keystrokes silently dropped on macOS; screenshots work, permissions verified good
- [BUG] Claude Desktop macOS: MCP server child process cannot obtain Calendar (EventKit) TCC permission — Reminders works, Calendar doesn't
- Add Slovak (sk-SK) language support for voice dictation
- [BUG] Custom URL scheme links (message://) not opened by OS in Claude Desktop — same root cause as #26952
- [Bug] Structured output generation fails after substantial token usage
- [Bug] Claude Code loops on same response ignoring user rejections
- [Bug] Excessive token consumption on simple tasks causing unexpected API costs
- [Bug] /rename command fails to generate accurate session names based on content
- Extension causes high cpu load
- [Bug] Code review triggers API content policy safeguards
- [Bug] Claude Code: Extremely low task completion rate and execution failures
- [Bug] Tool execution performance degradation and unreliable behavior
- [Bug] Tool execution loss of work and state management
- [Bug] False positive cybersecurity flag on test framework code
- [Bug] CLI not updating or showing update progress
- [Bug] Claude Code unable to complete complex multi-step code generation tasks - CHAT CGPT CHAT BOT DID IT IN 2 SEC
- [Bug] Legitimate request incorrectly flagged as security violation
- [Bug] Safety Guidelines Blocking Legitimate Cybersecurity Work
- [Bug] Security flag causing application freeze/hang
- [Bug] Unexpected PR comments without user authorization
- [Bug] Playwright test flagged by Fable analysis
- [Bug] Credits depleted without proper usage accounting or billing when exceeding monthly max limit
- I don't have access to the specific bug report details associated with this request ID. To generate an appropriate GitHub issue title, I would need you to provide: 1. A description of the bug or issue encountered 2. Any error messages received 3. The step
- I appreciate you providing context, but I need the actual bug report or issue description to generate an appropriate GitHub issue title. Could you please share: 1. What problem you're experiencing with Claude Code 2. What error messages you're seeing (if
- [Bug] CLI session expires and requires daily re-authentication
- Fable API error for safeguards on a legitimate and safe local single dev AI project
- [plugin: telegram] server.ts spins at 100% CPU when multiple Claude Code sessions run concurrently — PID_FILE assumes single writer
- telegram plugin: infinite CPU loop on EPIPE (broken pipe causes uncaughtException storm)
- [BUG] Race condition when pressing up arrow key
- MCP stdio servers lose stdin pipe after ~10-20 minutes, process orphaned to PID 1
- Suggestion: add "verify existing state before coding a fix" discipline to planning phase
- Monitor tool: completion notification silently dropped (nondeterministic) when a polling until-loop Monitor finishes (Windows)
- [BUG] `claude -r` doesn't find sessions when ~/.claude/projects/ entries are symlinks
- [Bug] False-positive Usage Policy block on legitimate TB epidemiology literature review
- [BUG] Cowork VM timeout after 60s on Windows 11 Pro ARM64 - stuck at add_plan9_shares
- [Bug] Claude Code violates architectural and testing rules under task pressure; inlines SQL, skips HTTP/JWT tests, ignores migration init patterns
- [Bug] Missing bug report content
- Bug: 'Failed with non-blocking status code: /bin/sh: python: command not found'
- [BUG] in the CLI, the text box caret has started to flash at epilepsy speed
- Custom agents created during session not discoverable until restart — request hot-reload
- Allow programmatic renaming of the VS Code extension session tab (skill/setting/API)
- Feature request: Sync ~/.claude/ setup (skills, plugins, MCP, statusline) across machines via Anthropic account
- BUG] macOS Quick Entry voice: Enter pressed during recognizer warmup strands transcript in text buffer without raising the window
- Cowork — all tasks stuck "running" (locked composer) after hard reboot
- [BUG] Extension shows "usage limit reached" while /usage reports 17% session / 1% weekly
- [BUG] Remote Control mobile: multiple permission prompts only show first, rest cause hang
- [BUG]
- [BUG] VS Code extension: deleting a conversation from history does not terminate its stdio MCP server (process orphans until full window restart)
- Allow suppressing/theming the interactive startup welcome (logo · version · "Feature of the week")
- [BUG] Claude Code Authentication Broken on Android + Codespaces (OAuth Redirect Issue)
- Claude Sonnet 4.6 – Deployed 56-agent parallel for simple image upload task
- [Bug] Excessive agent spawning causes token bloat for simple code analysis tasks
- System prompt hardcodes stale user path (-home-e) and writes to disk without user consent
- Claude Code VS Code extension forces 1M context on Pro plan - Windows
- [BUG][Windows][native-binary] v2.1.159+ silent-exits during startup — v2.1.112 unaffected
- [BUG/UX] Custom user command silently shadows built-in alias with no warning at creation or invocation time
- `/deep-research`: `[verified]` tags aren't bound to a primary-source check — checkable facts can pass on inter-model agreement / secondary sources
- [BUG] Remote Control: MCP permission prompts for non-read tools never surface in claude.ai/code web UI