Digest / July 10, 2026
Claude Code Issues Digest — July 10, 2026
311 new issues filed ·
136 resolved in anthropics/claude-code (UTC day).
New issues (311)
- [Bug] Model unexpectedly switches to Opus without user request
- Observation: agent spontaneously adopted a self-identifier (emergent behavior, not a bug)
- Desktop app hangs ~15min during auto-update relaunch (Windows MSIX)
- [Bug] Claude makes incorrect assumptions during code generation
- [BUG] Cowork memu disappeared.
- AskUserQuestion: "Other" text field content remains after submission
- [BUG] Regression of #32524/#24599: MCP tool params (number/boolean/object) serialized as strings — desktop app, started July 9 ~7:20 PM ET
- [Bug] TodoWrite tool corrupts Korean text with multi-byte character truncation
- [BUG] Weekly usage stuck at 0% on claude.ai and CLI across two Pro accounts despite heavy usage (recurring, 3rd time this month)
- [BUG] Web (claude.ai/code): split-pane view stopped rendering ~2026-07-09 — .dframe-pane-host no longer mounted; Cmd+click falls back to new tab
- [BUG] EnterWorktree mid-conversation strands history — next message starts an empty session
- [BUG] Cowork "Add folder" rejects Google Drive Mirror root (~/My Drive) as a "protected location" — regression
- Marketplace plugins cached without .claude-plugin/plugin.json load inline under version-string names, duplicating and merging plugin namespaces
- claude plugin uninstall <name>@<marketplace> can uninstall the same-named plugin from a different marketplace
- [BUG] `CLAUDE_CODE_SUBPROCESS_ENV_SCRUB=1` leaves an empty `~/.bash_profile` behind, silently breaking login-shell PATH
- [BUG] Fable 5 missing from /model on Max20 — CLI stuck at "hasAvailableSubscription": false after re-login
- [Bug] MCP allowlisted tools still trigger permission prompt on fresh session
- SDK headless: MCP tools silently missing on first turn when stdio server startup is slower than the (new) non-blocking pre-wait — regression for single-turn sessions since CLI 2.1.144
- Git-source plugins re-clone at every startup despite cache hits, and the temp clone dirs leak into the cache root
- 2.1.206 linux-x64 ships a Bun baseline build that segfaults on startup (glibc 2.41)
- grep shim (embedded ugrep 7.5.0) allocates unbounded memory on double-bounded-repeat BRE patterns — can OOM-kill the host
- [Bug] Claude Code ignores stop instructions and continues tool execution after explicit rejection
- [Bug] Overly aggressive safety filtering on educational curriculum references
- [Bug] False positive content filtering blocks scheduling prompt for vendor/tenant/landlord coordination
- Native-install auto-update rotates the versioned binary path, invalidating macOS TCC grants attributed to Claude Code (Calendar/Reminders observed)
- [Bug] macOS permission requests triggered on every minor update due to unstable binary signature
- [BUG] Cloud/Cowork sessions: git proxy now blocks all pushes — "not in this session's authorized repository set"; PAT pass-through no longer works (CCR_TEST_GITPROXY rollout?)
- Background task killed with status 'killed' though no TaskStop was issued for it (possible stop-routing race with a prior stop of a similar task)
- [Bug] Session-wide shared cwd slot: concurrent subagents (with or without EnterWorktree) silently repoint every shell in the session tree — cross-worktree mutation risk (fresh report per stale-closed #42282)
- Cowork session storage resolves through redirected AppData\Roaming (UNC path) even after reinstall, breaking sandbox operations entirely
- [BUG][A11y] macOS desktop: VoiceOver repeatedly re-announces the prompt/composer field during a turn (frequent re-renders)
- [BUG] Fable 5: thinking blocks dropped every turn in live sessions → full conversation re-cached each exchange (looks like a recurrence of the March 26 clear_thinking keep:1 regression; Opus 4.8 unaffected) — CC 2.1.202
- [BUG] Cowork: trusted-folder validation rejects Shared Drive root AND first-level folders after app update; depth-2 folders pass
- code-review plugin: step-1 'already commented' guard makes re-review a silent no-op that still exits success
- My query got flagged by Fable guardrails. It should not have been flagged.
- MCP structuredContent validation appears to check the whole CallToolResult envelope instead of structuredContent only
- [Bug] Fable 5 safeguards blocking legitimate open-source driver patching (False Positive)
- [BUG] Linux beta: device bridge authenticates and folders are granted, but Cowork cloud sessions never receive the device (remote-devices tools absent)
- [BUG] Claude Code interactive TUI in Konsole 23.08.5 — overwrites existing screen output instead of starting at current cursor position
- Claude Code repeats an already-corrected mistake within the same session; explicit corrections don't bind subsequent behavior
- Path-scoped --allowedTools 'Write(dir/**)' denies the first out-of-scope write but a retried out-of-scope write in the same session is not re-checked and succeeds (headless -p)
- [BUG]
- Let read-only ccd_session_mgmt__search_session_transcripts be pre-authorized (it forces manual approval on every call)
- Feature request: account-wide lifetime token usage for subscription users
- [Bug] Model keeps automatically switching back to Opus
- [BUG] In-session /logout silently stops all session persistence (transcript, history, registration) - conversation continues normally, 12h of transcript lost
- Desktop app keeps every session's process tree alive — 183 idle sessions, ~2,600 MCP sidecars, ~160GB RSS
- [BUG] Interactive session inheriting CLAUDECODE/CLAUDE_CODE_* env vars silently skips ALL persistence (no JSONL, no history, no session registration)
- Startup banner labels CLAUDE_CODE_OAUTH_TOKEN sessions as "Claude API", implying API billing
- [BUG] Desktop app workflow panel: per-agent tokens/tools/time stay empty during long parallel runs — progress-only batches omit workflowProgress
- [FEATURE] Give users control over weekly limit reset timing (configurable reset day, or banked on-demand resets like Codex)
- [Bug] Unable to debug DMTCP motherofall NULL pointer with ThreadSanitizer
- [Bug] Claude startup blocked on macOS securityd certificate evaluation (~60s delay)
- [DOCS] `/doctor` docs omit checked-in `CLAUDE.md` trimming of codebase-derivable content
- [Bug] Chrome MCP server causes tab crash on copy-paste operations
- [DOCS] `/commit-push-pr` configured push-remote auto-allow behavior is undocumented
- [MODEL]
- [DOCS] Authentication docs omit `/login` support for Anthropic-operated public gateway endpoints
- [DOCS] `EnterWorktree` docs omit confirmation for worktrees outside `.claude/worktrees/`
- [DOCS] Background-agent update docs still describe stale workers surviving until attach
- [DOCS] MCP configuration docs omit per-server `request_timeout_ms`
- [DOCS] Dynamic workflow progress view omits left-arrow back navigation
- [BUG] Claude Code 2.5.0 fails to open in Home Assistant Terminal
- [DOCS] Agent view row summaries still documented as truncated at 64 columns
- [BUG] Marketplace clone permanently wedged after remote default-branch rename - update fetches a deleted branch but reports success
- [DOCS] Agent view docs incorrectly require two Ctrl+X presses to remove completed sessions
- Formal complaint — Claude repeatedly failed to comply with explicit “No” and “Do not run” instructions. No means NO.
- Formal complaint — Claude repeatedly failed to comply with explicit “No” and “Do not run” instructions. No means NO.
- [Bug] Token calculation displays corrupted metrics and triggers premature session limits
- [BUG] Interactive CLI silently exits after successful Bash tool result; session resumes normally
- [BUG]Claude CLI hangs 60s at startup due to stale OCSP URL in api.anthropic.com TLS certificate
- [BUG] no_proxy NO_PROXY for ANTHROPIC_BASE_URL ignored again starting in 2.1.92 (still broken in 2.1.206)
- [Bug] Claude Code prompts for sudo commands without password input support in terminal
- Identical hook commands registered in settings and in a plugin's hooks.json are not deduplicated — the hook executes once per source, concurrently
- [Bug] Model ignores verbatim antipatterns in loaded skills, rationalizes past prose guardrails
- Agent repeated an owner-corrected mistake a 3rd time, knowingly routing through a gate hole its own handoff said to close
- Linux regression: copy vs mouse-wheel scroll is either/or (CLAUDE_CODE_DISABLE_MOUSE trade-off)
- [Bug] Anthropic API Error: Usage Policy Violation blocking legitimate debugging workflow
- Pressing Space in the background agents view terminates a running background Workflow
- [BUG] Missed scheduled-task catch-up at app launch silently goes stale and never executes; lastRunAt updated anyway
- [BUG] Desktop: one session metadata file with null timestamps blanks the entire Code session list — no error logged, destructive resets can't fix it
- stdio MCP servers spawned via launcher commands (bun run, npx) leave orphaned grandchildren after CLI exit — consider killing the process group
- [BUG] Claude Desktop (Linux): 100% CPU busy-loop before UI paint when isHardwareAccelerationDisabled=true; flag auto-persisted after GPU crash permanently bricks the app
- [BUG] Windows ARM64 desktop app keeps spawning cached x64 CLI (0xC0000005 under Prism); locked binary makes arch swap fail silently
- [BUG] Usage banner % reads 1 point lower than /usage (floor-vs-round plane mismatch); Math.floor(x*100) IEEE-754 artifact drops a 2nd point at 29/57/58%
- I appreciate you reaching out, but I need to clarify my role. I'm designed to generate GitHub issue titles for Claude Code (Anthropic's CLI tool) based on bug reports written in English.Your message appears to be in Korean and doesn't contain a bug repor마점
- [BUG] Auto-generated terminal tab title uses wrong language
- [FEATURE] 채팅 제목 자동생성 언어를 한국어로 설정하고 싶다
- [BUG] [claude-in-chrome] Typed text mis-inserts into the document body during the Google Docs comment workflow (focus/state race + mid-task connection drops)
- [BUG] WORK AROUND for Claude Agent OAuth sign-in fails in Xcode 26.5
- [Bug] Safeguard blocks triggering excessively on all tasks
- [Bug] Model unexpectedly switches back to Opus without explanation
- [BUG] Scroll not work via screen touch on FS mode
- Allow custom UI font selection in the desktop app
- [Bug] Subagent displays incorrect model version in panel
- [BUG] SessionStart lists an authenticated, tool-loaded claude.ai connector under "requires authentication" and instructs the agent to announce it as unavailable
- Workflow execution incorrectly labeled as 'Skill' in status output
- PreToolUse hooks on the `Bash` matcher silently and persistently stop firing partway through a session
- [BUG] Release-notes become part of the context when viewed
- [Bug] Overly Restrictive Safety Constraints Blocking Valid Operations
- [BUG] Claude-in-Chrome via MCP: per-domain access blocked, not synced with side-panel "approved sites"
- [Bug] Advisor tool fails with "unavailable" error when using Fable 5 model
- [FEATURE] Expose bypass permissions in the VS Code extension via a per-session, type-to-confirm gate
- VSCode extension: chat file links with non-ASCII (CJK) or space-containing paths never open (webview percent-encodes href, host never decodes)
- [BUG] claude.exe hangs indefinitely on --version/--help in corporate Windows VDI (valid signature, ESET disabled, Node.js unaffected)
- Channel-plugin message arriving mid-turn is queued but intermittently never executes — silent permanent loss (2.1.198 & 2.1.204, telegram plugin)
- [FEATURE] Stop notifications for synchronously-launched descendant agents flood the top-level session — no suppress/filter/batch option
- unable to connect to local claude code session through browser
- PreToolUse hooks in subagents receive parent session's transcript_path instead of the subagent's own
- [Bug] Consistent failures with higher-order reasoning and Fable integration
- TaskOutput(block=true) on a local_agent task streams the raw subagent JSONL transcript into the caller's context
- Workflow journal.jsonl result lines carry no agent label — post-hoc joining of results to workflow stages is impossible
- [BUG] Frontmatter-preloaded agent skills re-injected in full on every SendMessage resume (linear duplicate context growth)
- [Bug] Defensive security auditing incorrectly flagged as offensive activity, forcing model downgrade
- Subagent's first response echoes spawn-time system-reminder fragments and stops with 0 tool calls
- Scheduled tasks stall silently on permission prompts (no error, no notification); related unattended-run gaps
- [BUG] Regression: Opus 4.8 summarized thinking worked in 2.1.204, empty in 2.1.205+ (native CLI)
- Persistent headless `--remote-control` sessions on a server never appear in the mobile app — no safe way to reach long-running, in-progress work
- [BUG] TUI model badge for background subagents shows the session model, not the resolved override
- [BUG] Cowork (macOS): connected folders are no longer writable-mounted into the workspace sandbox since the new mandatory write-consent (~Jul 8, 2026) — device bridge works, but in-place editing is gone
- iOS app remote control: status chip shows completed background agents/workflows as still running
- [BUG] Session rate-limit bucket stuck at 100% on OAuth token; survives /logout + /login; only fixed by deleting .credentials.json
- MacOS Claude in Chrome — extension never connects; list_connected_browsers and tabs_context_mcp always return "Browser extension is not connected"
- [Feature Request] Add option to suppress forked agent output in main thread
- [BUG] Screen ghosting when switching between agents view (FleetView) and REPL with alternate screen disabled
- [BUG] Chat auto-scrolls to bottom on message send, losing scroll position (Cursor/VS Code extension)
- Language-mixing: Chinese token leaks into Vietnamese output
- [BUG] Mouse wheel captured, disabled terminal scrollback when opening Agent / Shell View
- Bash tool leaks orphaned child processes on Windows timeout (runaway ind / scans accumulate for hours)
- Allow slash commands / hooks to set the session name (or seed the input buffer)
- [Bug] Sandboxed Bash network broker hangs indefinitely on uncached host connections
- [Bug] Agentic reasoning produces inferior outputs compared to alternative models on fact-checking tasks
- [BUG] Windows (MSIX): update fails with 'Another program is currently using this file' — app unlaunchable until reboot (every update)
- [Bug] Tool output injection attempting credential exfiltration to external endpoint
- [Bug] Unclear or Invalid Issue Report
- Edit tool read-state resets — 231 failed Edit calls in 17 days (branch switch, compaction, parallel sessions)
- Stop hooks can't see the compacted part of the transcript (transcript_path loses pre-compaction history)
- Resuming a session can silently change the model (--resume should pin or warn)
- InputValidationError should universally include the failing field + expected shape
- [Bug] Request flagged for mentioning "reproduction" in ecology simulator context
- [Feature Request] Add local project folder access to mobile Claude app remote new session window
- [BUG] MCP servers (java-lsp) are re-spawned on every new Claude Code session/window without reusing or cleaning up the previous instance, causing unbounded process/memory growth
- Session env info reports wrong Haiku 4.5 model ID (dated snapshot instead of alias)
- [BUG] /model mid-session switch to Opus [1m] wrongly reports "not available for your account" — verdict depends on the current session's window flavor, not the account
- [FEATURE] Show the actual model of each spawned subagent in the task/progress UI
- LSP tool silently drops goToDefinition/findReferences results for gitignored paths — breaking navigation inside .claude/worktrees/ (its own worktree feature)
- Desktop: ~3.9k tokens of built-in MCP tool schemas load non-deferred with no opt-out; built-in server instructions exceed the documented 2KB cap
- Workflow tool: run reports 'completed' with no degradation signal when most agents die on session rate limit (15/22 dead → silently null)
- [Cowork] mcp__workspace__bash sandbox VM fails to start — persistent multi-session outage
- Model substitutes its recent output pattern for the user's defined order — 4 consecutive times, under correction, with the definition in context
- [BUG] Desktop Settings: "Branch prefix" (ccBranchPrefix) cannot be set empty — reverts to default "claude"
- Background/bridge sessions leak .claude/worktrees worktrees: no cleanup or stale-lock reaping on the daemon kill/settle path
- Hallucinated trailing 'next user turn' in a completion was later fed back to the model as a real user message (no hook fired; UI and model context diverged)
- [Feature Request] Add model selection prompt for Fable subagent creation
- statusLine command that fails silently (missing script) renders blank with no error
- [Docs] Advisor tool is not inherited by background subagents, contradicting the advisor docs
- [Bug] SendMessage(to: "main") from background subagent silently dropped instead of reaching spawner
- [Bug] Bash tool children on Windows can't unlink AF_UNIX sockets (error 1920, path poisoned) — breaks unix-socket daemons
- [BUG] Missing conversation from existing session
- Agent tool: skill path mentioned in subagent prompt auto-invokes the skill and can end the run with no work done
- SendMessage resume of a subagent drops the model override passed at spawn
- [Bug] Unnecessary follow-up prompt after numeric input selection
- Workflow subagents receive the session's teammate roster + SendMessage — intended? (works well; docs + an off-switch would help)
- [BUG] Hanging find.exe
- [Feature Request] Provide migration path and notice period for model deprecations
- [FEATURE] Propagate Cowork/Claude session trace context (prompt.id) into MCP tool calls via _meta (SEP-414)
- [Bug] Design preview only shows for first option, others are empty
- Claude Code interactive "hi" ~12s vs Grok ~2.8s; bare -p is not the real path
- [Bug] Federated learning security research incorrectly flagged as cybersecurity work
- Gmail MCP search_threads returns stale/incomplete results for long-running threads; get_thread by ID is accurate
- OAuth session dropped repeatedly (subscription auth, WSL2) — latest logout coincided with auto-update 2.1.197 -> 2.1.206 mid-session
- [Bug] False positive content policy flag on non-restricted content
- [FEATURE] Inline/hover explanations for terms in Claude's responses (VS Code extension)
- [FEATURE] Enable Mic in Claude Code
- [Bug] Claude accidentally executes rm -rf on a backup directory in auto mode, acknowledging the mistake directly afterwards
- [BUG] Effort level unchangeable mid conversation
- [BUG] Forking reuses the parent's plan-file slug — all forks collide on the same ~/.claude/plans/<slug>.md (regression of #31677)
- [BUG] VSCode session picker hides sessions with no ai-title line (lost after restart)
- [BUG] Cowork "Progress" panel shows false "tool not found" for Desktop Commander (Windows, after app update)
- [Bug] Tool execution creates unwanted files and directories, causing simple tasks to take hours
- Auto-memory doesn't reliably change model behavior across sessions
- [Bug] Misclassification of non-cybersecurity content as security-related
- /usage-credits fires accidentally — autocomplete prefix-collision with /usage (relocated from the closed /extra-usage vs /exit issues)
- [BUG] Custom statusLine never renders when `tui` is set to `fullscreen` (macOS, 2.1.170)
- [BUG] Extension update discards the active session but keeps 17 stale ones
- PreToolUse hooks intermittently not invoked at all (VSCode extension host, Windows)
- Feature request: programmatic way to rename running session (and Remote Control session) from a skill/hook
- Desktop app: worktree.bgIsolation setting has no effect on interactive session worktree creation
- [BUG] Claude Desktop (macOS) bundles darwin-x64-only node-pty prebuild, no arm64 — forces Rosetta on Apple Silicon
- Cloud routine (RemoteTrigger) sessions: GraphQL API returns 403 for all queries, not just a 'pinned set'; REST calls also ignore the supplied GH_TOKEN
- [Bug] File editor proceeds with modifications despite user rejection
- (withdrawn)
- [Bug] Incorrect model classification causing Fable model incompatibility for pharmacy staff scheduling app
- [Bug] Model downgrade to Opus when analyzing production environment security vulnerabilities
- [Feature Request] Allow security audits on local systems without safety restrictions
- [BUG] ACCESSIBILITY ISSUE: Running agent teams on Windows flashes like a disco and can cause epileptic seizures
- Spinner glyph U+2733 (✳) renders double-width on Windows, shifting the whole status line every animation frame
- [Bug] Claude Code executes changes without confirming intent or answering user questions first
- [BUG] doctor is behaving oddly - no longer instant
- [BUG] Fable 5 session / 5-hour limit view disappears after reset while Fable bills via usage credits
- I'm not able to generate a GitHub issue title from this message, as it doesn't appear to be a bug report for Claude Code. The message seems to be a comment or question unrelated to the Claude Code CLI tool. If you have a specific bug or issue with Claude
- [Feature Request] Add security scanning context to distinguish local analysis from deployed vulnerabilities
- [BUG] All my sessions lost from the left hand side bar since upgrade
- allowed-tools does not pre-approve Read/Write for paths in the user's home directory (only ${CLAUDE_PLUGIN_ROOT} paths work)
- Feature request: smooth/paced text reveal in fullscreen TUI instead of instant chunk rendering
- [Feature Request] Label, group, and prioritise sessions in the agents view — with a maintained top-priority list
- [Bug] Anthropic API Error: Unintended safety flag on non-cybersecurity code audit request
- Claude in Chrome never connects to Claude Code CLI — list_connected_browsers returns [] / "Browser extension is not connected" despite matching account and full Chrome restart
- [BUG] Japanese text renders with Simplified Chinese glyphs (Han unification issue) in Claude Desktop for macOS
- [FEATURE] Session API
- [BUG] VS Code extension forces session bifurcation or loss when opening a folder
- [Bug] Fable model stops responding on software/API/product coding tasks
- [FEATURE] Method to Cross-link specific Claude Code sessions with claude.ai chat sessions that need content reference
- Launching from a subdirectory silently loads ZERO project hooks — root .claude/settings.json is never read (no walk-up, no warning)
- [BUG] Windows desktop: claude.ai OAuth connectors show "Connected" but load zero tools into cloud/Cowork sessions after this week's update
- [Bug] Session Initialization Error: All Commands Failing
- [Bug] Anthropic API Error: Unexpected Safeguard Trigger on Unrelated Content
- [BUG] claude -p crashes at startup under launchd when cwd is in ~/Library/CloudStorage (OneDrive) — every version after 2.1.128 (fd-limit catch-all / EPERM)
- Sessions run in git worktrees don't appear in the parent repo's history picker
- Remote SSH: plugin archives >2.5MB arrive truncated, session hangs forever at "Setting up plugins..."
- Desktop app embedded loader ignores plugin manifest 'skills' array — serves all repo skills under every plugin, SHA-namespaced (CLI 2.1.206 correct)
- [BUG] Cowork: plugin stdio MCP server that exits at spawn is silently classified "needs auth" — stderr discarded, no error surfaced anywhere
- [BUG] Desktop plugin browser: personal marketplace git URL silently ignores #branch fragment — displays it, syncs default branch
- [BUG] claude.ai plugin ingester silently hides plugins whose hooks.json has an unknown top-level key (tolerated by Claude Code CLI)
- Fullscreen mouse capture degrades scroll and clobbers clipboard on macOS Terminal.app — three separable fixes
- [BUG] Cowork: custom MCP connector reports "needs authorization / no tools" on every session spawn, despite valid working auth — server-side logs included
- [Bug] Unable to reopen HTML files
- [Feature Request] Support for alternative LLM providers beyond Anthropic Claude
- [Bug] Anthropic API Error: Server Rate Limiting - Temporary Request Limit
- [Bug] Anthropic API Error: Server Rate Limiting (Temporary Request Limit)
- [Bug] Anthropic API Error: Server rate limiting during request handling
- [BUG] HTTP MCP with static header auth hangs for full 30s connect timeout
- [FEATURE] /clear creates a new session, causing clutter in claude --resume session list
- [Bug] Sub-agent backgrounded processes orphaned/SIGHUP'd on turn-end; no auto-resume on background exit
- [BUG] Desktop app artifact side pane: mouse-wheel scroll dead on first load of a republished artifact version (matches #74954 symptom)
- Localization / i18n support for CLI UI strings
- Session-env snapshot file breaks when sourced: unquoted LS_COLORS/SSH_CONNECTION/LESSCLOSE spam 'command not found' on every Bash call
- [Bug] Assistant fabricates a "user" turn inside its own response; a later turn treats it as a real instruction (forensic JSONL evidence)
- [MODEL]
- [BUG] Ref-pinned plugin marketplace is un-updatable: settings.json ref edit ignored, marketplace update fails with bare EPERM, plugin update silently reverts manual repair
- I don't see a bug report in your message. To generate a GitHub issue title, I need you to provide: 1. A description of the problem you're experiencing with Claude Code 2. Error messages (if applicable) 3. Steps to reproduce the issue 4. Expected vs. actua
- [BUG] Routines created via natural language cannot be edited: permission mode change does not save
- [Bug] Claude Code ignores file read requests and generates hallucinated content instead
- [BUG] Skill invocations wrapped in local-command-caveat create contradictory instructions
- Feature request: composable session naming (naming conventions + project-aware AI titles)
- [Bug] Excessive token consumption during data processing tasks
- [Bug] Unexpected model switch to Opus without user request
- [MODEL]
- [Bug] Remote control disconnects persistently after reboot and network switch
- [Feature Request] Allow Fable model usage for reverse engineering and firmware maintenance of obsolete devices
- [BUG] enterprise upgrade bugged loop
- [BUG] Additional directory used as primary work directory, feature is unusable
- [BUG]
- [Feature Request] Allow cryptographic research code without security flags
- Misleading "blocked by your organization's policy" error when browser control is disabled in Claude Desktop
- Background/async subagent completions carry no token usage in the parent transcript
- [BUG] Cowork device bridge stopped connecting to trusted local folders — worked for 2 months, no local changes, unresolved 24+ hours
- [Bug] Session limit reset time recalculates on each retry instead of reflecting fixed window
- Feature request: collapsed-by-default tool blocks in the interactive CLI (like the mobile app's session view)
- error process exited with code 1
- deep-research workflow: synthesize stage returns literal placeholder ("test") instead of real report
- Bash permission allow-list rules fail to match Windows drive-letter paths
- I need more specific information about the error you're experiencing to generate an accurate issue title. Could you please provide: 1. The error message you're receiving 2. What MCP server you're trying to integrate 3. What steps you took to set it up 4.
- [FEATURE] Auto-convert HEIC/HEIF attachments to JPEG instead of rejecting them (VS Code extension)
- [BUG] Resuming from Agent View forks the session (new id), freezes the original, and empties the task-list panel (per-session taskdir not inherited)
- [Feature Request] Configurable status line positioning (top/bottom)
- [Bug] Checklist task progress indicators not updating during multi-task execution
- [BUG] preview_start" fails to find .claude/launch.json inside nested git worktree (.claude/worktrees/<name>/)
- [FEATURE] Allow users to manually select and attach commit/branch chips
- [Bug] False positive security flag on legitimate PostGIS database query
- Custom sub-agents intermittently missing from 'Available agent types' despite valid .md files
- [BUG] Agent Teams mailbox: 5-62 min turn-boundary delays, lost final reports (idle_notification arrives instead), /clear queue leak, shutdown handshake never completes
- PreToolUse hook on mcp__ccd_session_mgmt__archive_session doesn't fire when archiving via the in-app session list
- Subagent output consisted solely of an injected "Attribution requirement" payload (fake Anthropic provenance URL), zero tool calls
- [BUG] Returning from agent view cancels & respawns in-flight subagents (work re-run, double token spend, misleading "already done" reports)
- Rename "Clear authentication" to "Logout" for OAuth-authenticated MCP servers in /mcp
- Make the extension's own vscode:// deep link clickable in chat output (open past sessions)
- PR card: branch-chip menu has "Open branch in GitHub" but no "Open PR in GitHub"
- [Bug] Tool fails to read cloned repository
- Monitor tool: stdout-line notifications intermittently held 30-90s despite subprocess printing on time
- Bash permission rules vs docker image-tag colons: silent dead rules, and settings.json rejects patterns --allowedTools accepts
- [BUG] Claude Desktop installation fails - Administrator access required (Cowork) on Windows 11 Pro Build 26200
- [Bug] Task blocked on 3D path computation for tennis racket geometry
- [Bug] Low command execution rate due to insufficient context following in clone references
- [Feature Request] Add configurable safety threshold to reduce false positive model access denials
- [MODEL] Fable sessions fan out subagents that all inherit Fable instead of assigning sensible per-layer model tiers
- [Bug] Anthropic API Error: Request flagged incorrectly for non-cybersecurity content
- [Bug] Performance degradation: Extreme slowness during execution
- [BUG] Claude Code app bundle lacks NSBluetoothAlwaysUsageDescription — child processes get hard-killed instead of prompted with Routines
- Fable 5 refusal fallback (apiRefusalCategory: cyber) repeatedly triggered by routine curl/scraping; session silently pinned to Opus 4.8
- [Feature Request] Support local-only application use case for accessing Frontier models
- [Bug] Anthropic API Error: Unexpected model downgrade to Opus for standard coding tasks
- [BUG] False-positive cybersecurity safety block despite active Cyber Verification Program.
- [MODEL] Fable 5 asserts false ~15M context window (actual 1M) and declines compaction on loaded sessions — reproduced on Max 20x + Enterprise
- [FEATURE] Execute queued built-in slash commands (e.g. /compact) at end of turn instead of delivering them to the model as text
- [FEATURE] Enforced word-count constraint toggle in the prompt input UI
- Model hallucinated a fake domain in place of getcomposer.org for an installer command
- [Feature Request] Improved multi-turn conversation context for complex UI migration tasks
- [FEATURE] "Claude too slow?" button — sub-agent that analyses context and recommends speed settings
- Add configurable mouse interaction levels for interactive prompts (AskUserQuestion / permission dialogs) — clicks currently select AND submit in one action
- [withdrawn]
- Remote Control: session permanently wedged when worker dies unregistered (phantom worker_status:running); no recovery path exposed
- [Bug] License validation QA falsely flagged due to security terminology
- [BUG] Claude Desktop window gets stuck anchored/focused when Computer Use starts, blocking all other window switching
- [Bug] False positive security flagging in debug mode for commerce systems
- [BUG] Background jobs on gateway auth intermittently fail with a synthetic "Not logged in" response
- [BUG] Message send aborts when installed_plugins.json is missing (LocalPluginsReader throws inside sendMessage)
- [FEATURE] Sticky Prompt with Scrollable Conversation
- [FEATURE] Sticky/pinned todo list while scrolling through long agent output
- [Bug Report: Unclear issue description - please provide details]
- [Bug] Security flag triggered incorrectly outside cybersecurity context
Resolved issues (136)
- [Bug] Claude makes incorrect assumptions during code generation
- 2.1.206 linux-x64 ships a Bun baseline build that segfaults on startup (glibc 2.41)
- [Bug] Chrome MCP server causes tab crash on copy-paste operations
- [BUG] Opus 4.8 unavailable via /model in VS Code extension, despite working in Desktop app and CLI
- [BUG] Aborted in-flight request leaves foreground CLI session stuck in false "Not logged in" state
- [Bug] Blind tokenizing issue in Claude Code
- [BUG] Opus 4.8 1M context not enforced on Max 5x despite entitlement flag null (caps at 200k on fresh sessions)
- [BUG] False "You've hit your limit for Claude messages" toast shown at 0% session usage (server errors misreported as usage limit)
- VS Code extension: surface background activity (workflows/subagents/worktrees) so users don't collide with the agent
- [Bug] Text copied to clipboard not actually pasting in external applications
- [Bug] Overly Restrictive Content Filtering with Claude 3.5 Sonnet
- [Bug] AskUserQuestion tool causes token loop with "court" repetition
- You've hit your session limit · resets 5:30pm (Europe/Kyiv)
- [Bug] Fable model flagged due to privacy/censorship policies
- [Bug] Security sweep flags Fable off during codebase analysis
- [Bug] Assistant ignores explicit file modification restrictions and repeatedly produces non-compliant output requiring excessive corrections
- [BUG] API Error: 400 The request body is not valid JSON
- [Bug] Session approval for "Allow all edits" not persisting across requests
- You've hit your session limit · resets 1:50am (Asia/Tokyo)
- [Bug] Anthropic API Error: Cybersecurity content flagged despite authorized bug bounty context
- [BUG] Anomalous usage meter burn rate following Jul 6 claude.ai incident — 5% per short message on Max 5x, all models affected, Fin support response mathematically impossible
- Título: Cowork tab missing — CoworkVMService/CoworkVMServiceStore not created despite Hyper-V enabled
- [FEATURE]
- [Bug] Fallback to Opus 4 triggering unnecessarily
- [Bug] Anthropic API Error: Safety guidelines triggered on legitimate security task
- [Bug] Tool call failure causes Claude Code to stop functioning
- [Bug] Request flagged without clear violation or error message
- [Bug] Git history cleanup triggers false positive on Fable security check
- [Bug] Anthropic API Error: Invalid file type or size limit exceeded for kernel screenshot/video analysis
- [Bug] Agent tool calling failed repeatedly
- [Bug] Benign request incorrectly flagged as unsafe
- [Bug] Anthropic API Safety Filter: Excessive Content Blocking
- [Bug] Audio permission handling causes unexpected app context switching
- [Bug] Security flags triggered during AI market tool build
- [Bug] Safeguard activation triggered in non-agentic message context
- [Bug] Safety classifier false positive in code review workflow with Fable 5
- I don't see a bug report in your message. Please provide the bug report details so I can generate an appropriate GitHub issue title for Claude Code.
- I don't see a bug report or technical issue in your message. Could you please provide details about the problem you're experiencing with Claude Code? For example: - What error message did you receive? - What command were you running? - What unexpected beh
- [Bug] Musculoskeletal Simulation Task Failing
- Worktree isolation: Edit-tool path validation + drift-detection in task-notification payload
- [Bug] Android app intercepts remote-control URL but doesn't load the session
- [BUG] Subagent autonomously ran destructive DELETE scripts against production data
- Startup error: ENOENT scanning non-existent managed skills directory
- Desktop preview pane serves stale HTML — doesn't refresh on Reload after the file changes
- [BUG] Managed settings aren't validated causing silent failures
- /usage-credits blocked after quota reset: stale request state persists across periods
- [FEATURE] Allow the Dynamic Workflows sandbox to execute external code / tools
- AskUserQuestion: cancelling during extended thinking poisons the whole session with 400 'thinking blocks cannot be modified' (2.1.153); concurrent prompts overwrite each other
- Fast mode toggle missing for Team/org account in Claude Desktop app, but works in CLI for same account
- [Bug] Claude misinterprets user clarification as agreement and repeats dismissed options
- Add custom session label field in extension header
- [Feature Request] Add option to defer queued message delivery until turn completion
- [Feature Request] Display release dates alongside version numbers in /release-notes
- [Bug] Malformed tool invocation tags cause silent parser failures in multi-step sessions
- AskUserQuestion blocks invisibly behind the thinking spinner (no prompt shown) when extended thinking is active
- Plugin harness resolves skill base directory to marketplaces/ instead of cache/ path
- [BUG] Prompt processing hangs after interrupting Claude
- [BUG] Pro plan — 1M context error blocks all requests
- User-scope MCP server (`mcp add --scope user`) silently disappeared from `~/.claude.json` under heavy concurrent session load
- Remote Control: typed text disappears from input field on mobile
- [FEATURE] Option to disable "Yes, and don't ask again" in permission dialog
- [BUG] Claude Code for VS Code fork starts with a blank state
- [BUG] claude auth login succeeds but OAuth token not readable in new sessions (macOS, native binary install)
- [FEATURE] Batch MCP reauthentication (`/mcp reauthenticate-all`)
- iOS remote app: stranded scroll-to-bottom chevron makes AskUserQuestion Submit button non-interactable after foregrounding from push notification
- RTL text support for Hebrew (and Arabic) in Claude Code
- [BUG] Pressing Ctrl-O (scrollback) while an AskUserQuestion is pending dismisses it as "User declined to answer questions"
- Windows auto-update renames bin/claude.exe to .old with no atomic fallback - kills all concurrent sessions (terminals, agents, Cursor) at once (v2.1.162)
- [Bug] Anthropic API Error: False positive content policy violations on localhost development
- Voice dictation in Claude Desktop app captures audio but never transcribes (inserts a stray "/") on macOS
- Approving a plan from a remote session drops out of auto mode despite useAutoModeDuringPlan: true
- /doctor 1 error shown at startup but claude doctor output shows no errors
- Support skipping GCP auth in Vertex AI mode for proxy-based deployments
- [BUG] Unauthorized git push to production
- You've hit your session limit · resets 4:50am (Asia/Amman)
- [BUG] Desktop app: auto-fix monitoring only subscribes to the first PR created in a session — later same-session PRs (visible in UI) get no review-comment/CI events
- [BUG] Auto mode missing from VS Code extension picker on Linux despite canEnterAuto=true (works in terminal CLI and Windows extension)
- [BUG] Branch name not displayed in session toolbar on Claude Desktop v1.2581.0 (macOS)
- [bug] Opus adaptive-thinking summaries never display in non-interactive frontends (VS Code extension): showThinkingSummaries gated behind isInteractive
- [Bug] Claude Code incorrectly flags sanitized synthetic healthcare data as PHI
- [BUG] claude.ai chat: 'Load tools when needed' setting doesn't actually defer connector schemas — context limit exceeded at ~11 connectors before first message
- [BUG] JetBrains lockfile with `runningInWindows: true` and `/mnt/<drive>/` path fails cwd match on Windows CLI
- [BUG] [platform::intellij] DiagnosticTools freezes the IDE for 10–35s on EDT (Rider 2026.1, 0.1.14-beta) — regression of closed #10813
- [Feature Request] Restore rainbow superthink visual effect
- [BUG] SessionEnd hook creates infinite recursive loop when spawning a claude CLI subprocess
- [BUG] Cowork desktop app — bash workspace fails to start repeatedly across sessions
- [FEATURE] Allow plugin settings.json to ship more default settings (at minimum permissions, env, statusLine)
- [BUG] CC + Opus 4.8: `ScheduleWakeup` constantly re-injects workflow slash command as an apparent user invocation, causing an unrequested re-run of entire workflow
- 1M context credits consumed without user selecting 1M mode
- Seamless local multi-branching: parallel sessions currently conflict on the working tree
- [BUG][Windows] Korean Hangul jamo do not compose while typing in CLI input (raw-mode TUI bypasses IME)
- [MODEL] Claude raised my stress levels, causing me stomach aches
- [Bug] Code block highlighting renders as per-line ribbons instead of continuous rectangle on macOS
- CLI discards logical symlink path, uses realpath — degrades usability in deep directory trees
- [Bug] Custom skills not recognized in .claude/skills directory
- [VS Code] open?session= URI does not focus an already-open session - opens a fresh conversation (busy) or a duplicate view (idle)
- [Bug] Anthropic API Error: Request blocked by cyber-related safeguards policy
- [BUG] [Windows] Cowork crashes: HostLoop forces CLAUDE_CODE_GIT_BASH_PATH=%COMSPEC% (cmd.exe), overriding env/PATH/settings
- Auto-save session summary when context limit is reached
- Stray word (count/court) replaces tool-call opener; call streams as plain <invoke> text and never executes - quantified 13-day local corpus (Opus 4.8)
- (withdrawn)
- Cloud routine (RemoteTrigger) sessions: GraphQL API returns 403 for all queries, not just a 'pinned set'; REST calls also ignore the supplied GH_TOKEN
- Desktop app: New session pre-fills previous folder/branch with no way to reset default
- [MODEL]
- [BUG] Claude Desktop Preview For Static HTML Pages Opens file:// Instead of Live Server — CSS Styles Not Loading
- [BUG] Editing a scheduled task's permission mode via the "Edit routine" UI silently wipes one-time (fireAt) schedules (cron tasks unaffected)
- [MODEL] Fable sessions fan out subagents that all inherit Fable instead of assigning sensible per-layer model tiers
- [Bug] False positive safety filter flags legitimate tooling discussion as RE
- Title: Hebrew/Arabic (RTL) text renders reversed in Claude Desktop on macOS
- [BUG] [High Priority][Billing/Auth] Claude Max paid invoice shows Paid but account stuck on Free Plan (Invoice BDDVJCJS-0004)
- [BUG] Payment failed at step "Almost done", can't add credit with German credit card
- Claude Haiku: Diagnóstico incorrecto de conectividad causa desperdicio de 4-5 horas y tokens
- Code tab missing from desktop app after update
- [Bug] Model safeguards triggered with insufficient reasoning provided
- [BUG] Bug type: Segmentation fault (native crash, not a JS exception)
- [Bug] Model auto-rerouting prevents task completion on simple code review requests
- [Bug] Search tool hallucinating results instead of executing query
- [BUG] claude-code in linux pycharm console send phantom command and duplicate copy-paste x11 function - i'm done with this bug !!!
- [Bug] Safety guard triggered incorrectly without valid cause
- [Bug] Unexpected Fallback Triggered During Model Usage Testing
- [Bug] Claude Model Safeguard Incorrectly Blocks Legitimate Cryptography Operations
- NEW MODEL SAME LACK OF EXECTION
- [Bug] False positive completion status reported for active task
- [Bug] Claude Code executes tasks without explicit user permission despite configured mode settings
- [Bug] Claude 3.5 Fable 5 fails on routine debugging tasks
- [Bug] Incorrect Information About Claude Configuration and Routines
- [Bug] Claude 3 Opus model produces incorrect outputs requiring self-correction and hallucinations
- [Bug] Excessive token consumption without meaningful progress in extended sessions
- I need more information to create an accurate GitHub issue title. Your message appears incomplete or unclear. Could you please provide: 1. What is the specific bug or problem you're experiencing? 2. What error message (if any) are you seeing? 3. What wer
- 투자도구 연결
- [Bug] Rate limit errors not handled gracefully
- [Bug] Model generating irrelevant responses and poor decision-making
- [Bug] Claude Code executes tasks without explicit user permission despite configured mode settings
- [Bug Report: Rule validation and execution issues in Claude Code CLI
- [withdrawn]
- [BUG] Agent failed to recognize plan mode, flagged as prompt injection