Digest / July 20, 2026
Claude Code Issues Digest — July 20, 2026
352 new issues filed ·
172 resolved in anthropics/claude-code (UTC day).
New issues (352)
- test
- [FEATURE] Allow "Last activity" time filter when Group by is set to Custom groups
- Opus 4.8 (direct API): thinking text empty (signature-only blocks) despite --thinking-display summarized — started 2026-07-19 ~13:07 UTC; Fable 5 unaffected
- [Bug] Anthropic API Error: 500 Internal Server Error
- [Bug] Anthropic API Error: 500 Internal Server Error
- [BUG] /config menu: Escape should cancel, not commit, the last-viewed setting
- I appreciate you sharing this, but I don't see a bug report in your message - you've only provided a file path to a markdown document on your system. To help you generate a proper GitHub issue title, please share the actual content of the bug report. You
- [Bug] Anthropic API Error: 500 Internal Server Error on specific project
- Monthly spend limit notification should include the session/usage reset time
- iOS app: every Remote Control send fails "Failed to send message - The request was invalid", while the web client works on the same session
- [Bug] Unable to Continue Project Creation
- I don't see a bug report in your message - you've only provided a request ID. Could you please share the actual bug report or error details so I can generate an appropriate GitHub issue title?
- Opus requests return 500 Internal Server Error in VS Code extension; Sonnet works fine on same session
- [Bug] Fable Safeguard incorrectly flags searchbar dictionary lookup
- Opus 4.8 fabricated attached-PDF content, embedding an unlabeled prompt-injection payload (fake exfil URL + concealment instruction)
- False flagged R code minification
- /mcp reconnect (all or single-server) always fails with "MCP controls aren't available right now" — persists across window reload, full VS Code restart, and OS reboot
- [FEATURE] Make background session idle timeout configurable
- Prompt injection appeared in subagent context; coincided with safety classifier being unavailable
- Cowork ARM64: stale VM instance holds smol-bin.vhdx (EBUSY) + zero HNS networks — recurring host BSODs (Samsung Galaxy Book4 Edge, Snapdragon X)
- [Bug] Unclear content policy flag on court/judgement search feature
- [Bug] Model downgrade from Claude Fable to Opus during security implementation tasks
- [Bug] Claude Code CLI exits with code 1 on Windows when processing image attachments
- You've hit your session limit · resets 10:20pm (America/Los_Angeles)
- /doctor on Windows: reports valid JSON as FAIL (jq missing) and can't find config when CLAUDE_CONFIG_DIR is set
- Telegram Channels: inbound messages never reach the session (outbound works)
- [Bug] Long file:// URLs broken by hard line wraps in Windows Terminal output
- [Bug] Claude Code caused system corruption during Debian installation with unsafe file modifications
- [Bug] Agent ignores specified task requirements and completes unrelated task variations
- pwned\
- [FEATURE] Agents view: mark the main session, and use color to keep parallel work legible in the terminal
- Any way to let Claude run /verify and /code-review on its own again?
- [BUG] Cowork browser picker shows generic "Browser 1/2/3" instead of configured browser names
- [BUG] - Localized footer (ar/hi/ur/fr) missing Services, Education, and Areas-We-Serve sections — ~65% fewer internal links than English on every non-English page
- Routines: model selector missing from create and edit forms — runs silently fall back to default model
- [Bug] Anthropic API Error: Slow Response and Persistent API Failures
- [BUG] VS Code editor focus/group behavior changed after Claude Code 2.1.215 update
- [Bug] Web scraper incorrectly flagged as cybersecurity task
- [BUG] Subprocess/binary hangs indefinitely on Windows — all diagnostics pass, no error ever returned
- allowedChannelPlugins in local managed settings silently shadowed by remote org managed settings; skip message points at the ignored file
- [Bug] Anthropic API Error: Invalid or Revoked API Key Authentication
- [BUG] Subagent SSE streams stall silently at scale; connections stay ESTABLISHED with zero inbound bytes (dynamic workflows, v2.1.214, Windows 11)
- [BUG] Model continues past end of turn, emitting a fabricated user turn and a fabricated system-reminder block inside its own assistant message
- I don't have enough information to generate a GitHub issue title from this message. The text "req_011CdCZxAL4bzcD7JYALctNh just trying to engineer a product" doesn't describe a specific bug or feature request. Please provide details about: - What problem
- Opus 4.6: Evidence-before-action violations and circular spec-fix regression across review rounds
- [Feature Request] Official support for Claude Desktop on Arch Linux and derivatives
- Windows: stdio MCP server spawning ssh.exe never connects (same server works via macOS client, Codex CLI, and manual pipe)
- Session .jsonl rewritten to metadata-only stub (zero user/assistant records) at resume time; subagent transcripts survive — 27 sessions affected, 2.1.21x
- [BUG] SessionStart hook stdout not injected into context (regression 2.1.209 -> 2.1.215, desktop app)
- Native Windows: working directory does not scope shell-tool filesystem access
- Native Windows: working directory does not scope shell-tool filesystem access
- [Bug] Cron jobs failing due to stale server references and stopped service dependencies
- Subagent `maxTurns` frontmatter is not honored (agent runs past the declared cap)
- [Feature Request] Restore discontinued Fable model support
- Desktop app: support custom themes / accent colors (parity with CLI custom theme system)
- [BUG] Scroll Lock key causes Claude Code to exit in VS Code integrated terminal (Windows 11)
- [BUG] Fresh install from claude.com/download resulted in Squirrel install, not MSIX — no errors shown, blocked Cowork
- [FABLE] Persistent multi-week inability to use Fable 5 for device/sensor-heavy (camera, mic, external-fetch) feature work — likely dual-use safety classifier over-triggering
- [BUG] Desktop "Open in terminal" says "session resumed" but creates a new session
- FUCK off anthropic you shit fuck heads I AM WORKING ON MY...model
- [FEATURE] VS Code extension: render MCP `claude/channel` push notifications in the native panel (parity with the terminal TUI)
- [BUG]Safety classifier false-positive on formal verification / EDA work
- [BUG] Claude in Chrome: phantom tabs accept and answer javascript_tool execution after Chrome has closed them
- I'm ready to generate a GitHub issue title for Claude Code. Please provide the bug report or issue description that you'd like me to create a title for.
- Desktop app: file:// markdown links open in in-app Browser panel even when 'Open links in Browser panel' is off
- Built-in Bash sandbox: network isolation unreliable on WSL2; failIfUnavailable doesn't detect a non-enforcing proxy
- [Bug] Main session persists and executes autonomously after /exit, ignoring CLAUDE.md directives
- [BUG] /hooks omits hooks from managed-settings.json even though they execute
- [BUG] Routine reports connectors as enabledInChat: false despite being connected globally and attached to the routine
- [Bug] Plan mode incorrectly requests permissions for non-destructive operations like explore and Bash commands
- [BUG] PreToolUse hook `updatedInput` is silently ignored for the Bash tool
- Agent View: show which sessions have an active Remote Control connection
- [BUG] Claude Desktop Mac Rendering Issue
- Routines: hourly cron minute is silently rewritten to the save-time minute (undocumented)
- [BUG] Interactive choice prompts (AskUserQuestion) show options but accept no keyboard input mid-session (macOS, Apple Terminal, v2.1.215)
- [Feature Request] Add option to disable response placeholder/prefix
- Bug - Les instructions d'une routine creee via l'API /v1/code/triggers ne sont pas executees
- Voice mode: captures good audio but transcription intermittently returns blank (WSL2/WSLg, v2.1.215)
- [BUG] "Claude is waiting for your input" notification fires on ordinary responses when window is unfocused, despite "Response completions" being off
- [FEATURE] Permission Dialog Position Conflicts with Windows Notifications
- [BUG] `ide_selection` broken when filename starts with square brackets
- [Bug] Claude Max subscription plan not recognized, charges usage credits instead
- I don't have a bug report to create a GitHub issue title from. Your message appears to be a personal note about backing up a website rather than a bug report for Claude Code. If you have a specific issue with Claude Code (such as an error message, unexpec
- [Bug] Credit Usage Mismatch: Showing Depletion Below 50% Usage on max(20x) Plan
- [Bug] Incorrect usage quota calculation for MAX 5X tier with Claude 5
- [Bug] Plan mode restriction not applied after model switch with /model command
- [BUG] Fable 5 prompts 'usage credits required' on Max plan from 2026-07-20, the day it became standard on Max (saved model claude-fable-5[1m])
- [Bug] Claude switching to Opus mid-session without user consent
- Model glitches mid-\uXXXX escape in tool-call arguments (emits \ + raw CJK char), making the whole input unparsable (InputValidationError)
- LLM systematically terminates verification scope via L4 reasoning — defects missed
- [Bug] Fable 5 incorrectly requires usage credits on Max 20x plan despite available weekly allowance
- Remote Control: reconnect burst wedges hosts in a no-backoff CPU spin (A); a busy turn is misread as "stopped responding mid-turn" (B)
- [Bug] Fable 5 unavailable despite unused allowance on Max plan
- [Bug] Fable model inaccessible despite included in max plan and available usage credits
- [Bug] Session rating registered on accidental number key press without confirmation
- [Annoying] /login flow is not smooth
- Monitor tool: background poll loop silently stops emitting output/heartbeats mid-run, with no error or completion notification
- [BUG] Outdated CLI (2.1.211) session hits spurious 'session limit' on Fable 5 and silently falls back to Opus 4.8 — no guidance to update/restart (works fine on 2.1.215)
- Feature request: set session color at launch (--session-color flag or sessionColor setting)
- Model selector reverts to Sonnet after "Fable 5 requires usage credits" dialog, despite Max plan including Fable 5 and `/model` reporting success
- [Bug] Display output not rendering
- [BUG] GitHub repos not visible in Claude Code cloud repo picker despite GitHub App installed and connector connected
- [BUG] Background shell task completes after agent moves on, silently overwrites files with stale content
- [Bug] Safety Check Triggered During NPU Access Verification
- PreToolUse "ask" permissionDecision (and permissions.ask list) not enforced — tools run without confirmation
- Sidebar & Cmd-K session list empties after app relaunch on macOS (data intact on disk)
- Desktop (Windows): 'Auto-fix CI and address comments' checkbox is a silent no-op since 1.22209.0 (worked in 1.21459.3)
- [BUG] Silent SIGSEGV (SI_KERNEL, addr=NULL) — CPU reports AVX2 but osxsave absent — regression 2.1.205→2.1.215
- [BUG] Fable 5 gated behind usage credits dialog on Max when authenticated via claude setup-token (inference-only scope cannot read entitlements)
- [FEATURE] Chips with dropdown functionality e.g. workfolder should have a chevron or other indication that they are dropdown
- [BUG] VS Code extension's native permission-prompt UI doesn't fire the Notification (permission_prompt) hook
- [BUG] Remote Control Model Selector Defaults to EXTRA effort with no changeability, overrides local CLI
- [Bug] Overflow Error in STM32 Pager Implementation
- [BUG] [Cowork] New project's Name and Description fields don't save — reverts to folder filepath, description blank
- Worktree sessions reuse an existing worktree directory from a previous session instead of creating a fresh one
- Docs mismatch: permission mode UI in Desktop app doesn't match documented 'mode selector next to send button'
- Voice dictation: "No audio detected from microphone" despite fully working mic (Arch Linux, PulseAudio)
- [BUG] Silent total transcript loss in stock macOS Terminal.app — 12h session wrote no .jsonl and no history entries, workflow sidecars persisted fine (same signature as #79188, no multiplexer)
- Auto-mode classifier too restrictive for private single-operator homelabs
- [Cowork] "Failed to load local file" is not txt/md-specific — `_readLocalHostFile` fails the same way for every file type, including xlsx/html which silently recover
- [Bug] Question by the agent not visible in dark theme
- [BUG] Remote-execution sandbox: LibreOffice cannot load any document — type detection fails for all formats, --convert-to exits 0 with no output
- [Bug] Unexpected token usage spike after weekly reset
- [BUG] Session silently stopped persisting to JSONL for 10 days while still working; resume after auto-update loads the stale state (Windows variant of #79298)
- [BUG] Project-specific Plugins/Skills show mismatched or duplicated data in Claude for Mac
- Claude Desktop (Windows) — steady kernel nonpaged-pool memory leak tied to app runtime, causes full system hang after ~1.5-2 hours
- "requires usage credits" error for Fable 5 despite all quotas well under limit (session 3%, week 26%, week Fable 24%)
- "Auto-fix CI & address comments" checkbox unclickable in PR pill
- "Check for updates" for local plugin marketplaces intermittently reports "Already up to date." when the repo is ahead
- [FEATURE] /nudge — send an ephemeral one-turn instruction that doesn't persist in conversation context
- Fabricated background-task result message appeared in conversation before genuine Workflow task-notification
- [BUG] Safety classifier intermittently unavailable, blocking tool execution in long sessions
- claude plugin list reports a plugin enabled with an installPath that no longer exists; payload is never re-materialized
- Claude Code (VS Code extension) prompts for usage credits on Fable 5 despite Max plan showing full entitlement
- Remote Control does not auto-reconnect after a network interruption
- VS Code extension: remoteControlAtStartup: true is not honored — sessions never auto-start Remote Control
- [Bug] Claude fails to execute single-level grammar task; requires extensive iterations
- [FEATURE] /export to export sessions in .html file
- [Bug] Safety policies incorrectly blocking non-security/biology open source work
- [BUG] Resource not accessible by integration" 403, GitHub
- [FEATURE] Configure list of session history to display in Claude desktop
- Workflow: schema-forced custom agentType can end its turn with prose and stall silently (no per-generation watchdog)
- Feature request: first-class flow for filing issues discovered mid-session (assistant-drafted, user-approved)
- Setting to hide/slim the input footer hints (mode indicator + "esc to interrupt")
- Visible context window usage indicator in Cowork[FEATURE]
- [BUG] No safeguard before an agent bulk-creates dozens of PRs against a real external repo (91 PRs auto-closed + repo locked me out)
- [Bug] Claude ignoring ruleset constraints and executing blocked commands without user consent
- I'm not able to generate an issue title from that input as it doesn't contain a bug report or feature request description. Please provide details about the actual problem you're experiencing with Claude Code, including: - What you were trying to do - What
- [Feature request] RTL support + word-boundary wrapping for Hebrew/Arabic in Claude Code (desktop app)
- VS Code extension /model toggle intermittently corrupts settings.json (malformed JSON), silently dropping all permissions and hooks
- [BUG] Regression: desktop sidebar "Last activity" filter hidden for all Group by modes except State
- [Bug] Onboarding guide sharing broken in 2.1.215: `check` crashes on e.length, `update`/`create` report success but write nothing
- [Bug] Tool ignores cloned repository context and requires repetitive instructions
- [BUG] macOS: locked login keychain silently breaks auth — /login falsely reports "Login successful", and entering the password at the keychain prompts still leaves the session logged out
- [Bug] Missing input validation for performance-critical operations
- [BUG] Cloud scheduled Cowork task completes and generates final message but hangs indefinitely at "Sending notification" — push/email never delivered
- Dispatch locks sessions to Fable 5 with no working model switch, blocking usage available on other models
- [FEATURE] Ability to revoke Claude's access to a specific tab without closing/reloading it
- /model shows Fable 5 as credits-only (€0.00) on Max plan, while /usage shows Fable covered by plan (5% of weekly limit)
- [BUG] Fable 5 usage-credits gate blocks session while weekly usage limit shows plenty of headroom (Pro plan)
- Plugin marketplace install fails on Windows when env.PATH is POSIX-style (git-bash format)
- [BUG] Background subagent turn produced corrupted non-text output, consuming full max_tokens budget
- [Bug] Claude Sonnet task execution quality degraded, model acting as executor instead of thinking partner
- [BUG] Teleport of a remote session fetches 0 log entries despite the session having a full transcript on claude.ai/code
- [Bug] Anthropic API Error: Invalid Tool Property Name Format
- [Bug] Usage token activation required despite valid max tokens and available quota
- Bridge/daemon-hosted sessions never write transcript to local JSONL; no way to re-host or export them locally
- [BUG] High memory usage during claude install.sh
- [BUG] Desktop Windows: global session list (sidebar Recents + Home "Active") is frozen at an arbitrary past cutoff, while the project-scoped list is correct — same app, same moment, two contradictory lists
- Claude Desktop (macOS): renderer EventEmitter leak (AutoUpdater/AppPreferences listeners) -> grey screen, dozens of restarts/day
- [BUG] .worktreeinclude patterns with leading `**/` silently match nothing during worktree copy
- Illegal instruction crash on ARM64 Linux in Apple Virtualization Framework (AVF) — embedded Bun runtime incompatible
- CLI silently ignores mistyped single-dash -dangerously-skip-permissions and runs fully prompted with no warning
- Shared claude daemon leaks ANTHROPIC_AUTH_TOKEN from first session into all later sessions on the machine — silent wrong-account auth/billing
- [FEATURE] UX for /workflow is missing some info / actual state of running task(s)
- [Bug] Anthropic API Error: Safety flag blocks legitimate game design content
- [Bug] Cross-session state inconsistency between distributed Claude Code instances
- [Bug] Conversation context lost or cleared unexpectedly
- [Bug] Claude Code creates feature branches that track protected branches, bypassing review gates
- Custom statusLine command still silently falls back to default in 2.1.215 (regression since 2.1.53, see #28526)
- [BUG] Cowork project groups disappeared after update (desktop + web), v1.22209.3
- [FEATURE] prompt for `cleanupPeriodDays` during initial setup
- [FEATURE] VSCode extension: render images inline in the chat panel instead of the "[Image]" placeholder
- [Bug] Anthropic API Error: 529 Overloaded
- Mobile Background-tasks panel shows completed subagents as Running with wall-clock timers; Stop button has no effect
- [BUG] Consantly getting: API Error: Stream idle timeout - no chunks received
- Bash tool shell runs with `expand_aliases` enabled — shell aliases can silently rewrite commands after PreToolUse hook approval
- [BUG] VS Code extension blocks Fable 5 with "requires usage credits" while account has 20% of weekly Fable allowance remaining (Max plan)
- [Bug] Tool blocked by system policy preventing execution and file operations
- [FEATURE] Interactive session deletion in the `--resume` picker (like Ctrl+X in `claude agents`)
- [Bug] Bug Report: Unable to improve security of user's own site
- Claude-in-Chrome: extension reports "not connected" persistently, even after Claude Code restart
- Linux desktop app: cloud (session_) chats never spawn a process → "send but no reply"; cowork VM "not supported (linux/x64)"
- [Bug] Missing explanations in Claude output display
- Desktop app: artifact-preview renderers leak ~1GB/hour, are never reaped, and survive app restarts
- ask PreToolUse hook decision silently fails to surface in a top-level session carrying CLAUDE_CODE_CHILD_SESSION=1 (command executes instead of pausing for approval)
- [BUG]
- [FEATURE] Allow Ctrl+D, Ctrl+D to exit `claude agents` (consistent with `claude --resume`)
- File viewer does not syntax-highlight Terraform (.tf) files
- [FEATURE] Add a setting to exit with a single Ctrl+D instead of pressing it twice
- [BUG]Claude in Chrome拡張機能：スクリーンショット/ページ読み取り系ツールが特定ページに依らず継続的にタイムアウトする
- [Windows] Navigation buttons overlap native window controls in RTL (Arabic) layout
- Agents view: @ project picker is prefix-only, so interior substring matches never surface
- `claude-api` skill dumps entire ~784KB SKILL.md into context for a simple usage-limit question, burning huge Pro-plan usage
- [BUG] [Billing] Pro→Max 5x upgrade blocked by Stripe Link error 3+ days, no card ever authorized — matches pattern in #56281, #55266, #57977
- Feature request: option to collapse/hide thinking blocks in the conversation view
- [Bug] Opus agent invocations charged despite available Fable usage credits
- Bash tool prompts for approval despite matching allow-list rule + defaultMode: bypassPermissions
- Credit balance is too low
- [Bug] claude-api skill consumes entire context window on Haiku model
- [BUG] Cowork (cloud) re-serves stale local-file content within a session: after a file is staged once, re-staging the same path returns the OLD bytes while reporting the file's NEW size/mtime (false all-clear)
- [Bug] Claude 3.5 Sonnet model access denied despite max plan subscription
- [FEATURE] Sort File Suggestions by Creation or Updated Date
- [BUG][Windows] Leftover `claude.exe.old.<ts>` process after auto-update enters a `taskkill` spawn loop, wedging WMI (`WmiPrvSE` ~400% CPU) machine-wide
- Keybindings: direct model-select and effort-adjust actions (not just open-picker)
- [Bug] Tool call occasionally emitted as literal text (missing namespace prefix, preceded by stray "card" token) instead of a real tool_use invocation
- [BUG] claude --resume silently returns a PARTIAL transcript: remote-control/phone turns are never written to local disk → overnight work lost with no error (+ sessions booted while credit-limited never register for remote control)
- [FEATURE] retrieve the search index's cached/extracted copy of a page
- [Bug] Anthropic API Error: Server Rate Limiting During Normal Usage
- [Bug] Anthropic API Error: Server rate limiting on valid requests
- [Bug] Anthropic API Error: Rate limiting - server temporarily limiting requests
- [Bug] Anthropic API Error: Server rate limiting on requests
- [Bug] Anthropic API Error: Server Rate Limiting During Request Execution
- Model selection scoped to a single task got applied to the whole session, causing unexpected token/credit consumption
- Session-wide model switch to Fable billed all turns at premium rate — no warning when scoping was intended to be per-task
- Feature request: Proactively suggest switching to a higher-tier model (e.g., Fable) based on task complexity
- PreToolUse hooks in project .claude/settings.json are silently not registered
- [Bug] Claude Code incorrectly gates Fable 5 behind usage credits for Team Premium seats
- /copy writes the clipboard as latin-1, so non-ASCII output pastes as mojibake (still broken in 2.1.215)
- [BUG] GitHub connector: repo selected via "Add from GitHub" never reaches the conversation
- Allow opt-in exemption for specific (write-capable) MCP tools in plan mode
- Emit OSC 8 hyperlinks for displayed file paths (like URLs)
- Option to paste an image file path as literal text (don't auto-attach)
- Mid-session plugin update leaves session in mixed state: new hook definitions resolve against the old ${CLAUDE_PLUGIN_ROOT} cache until restart
- [Bug] Cybersecurity Flags triggered when working on status website projects
- CLAUDE.md in --add-dir folders is silently skipped, including when Claude edits files in that folder
- [BUG] Claude Desktop MSIX install fails with HRESULT 0x80073CF9 due to CoworkVMService conflict (Windows 10 Pro 19045) — recurring after app hangs
- Desktop app (Code tab): failed claude.ai connector fetch cached indefinitely — connector silently missing from all sessions until manually toggled
- [BUG] Crash on session resume: "null is not an object (evaluating 'q.split')" in diff/edit renderer (regression of #38633)
- [BUG]: coordinator agent messages bypass plan-mode consent controls in sub-agents
- [Bug] System prompt injection in terminal interrupts user input and causes false feedback
- [BUG] Interactive sessions overwrite CLAUDE_CODE_TASK_LIST_ID with the project basename (regression 2.1.212 -> 2.1.214, still in 2.1.215)
- Mouse text selection + Cmd+C does not copy in the TUI (mouse mode captures the selection)
- Warn or confirm when a new session reuses the --name of another active session
- MCP server env ${VAR} left unexpanded when server is respawned mid-session (reconnect)
- macOS: recurring Gatekeeper "ripgrep.node could not be verified" when `claude -p` is spawned by a browser (native-messaging host) — quarantine-strip helper skips linker-signed modules
- [BUG] Headless `claude -p`: API errors (e.g. rate limit) return `subtype:"success"` + exit code 0 — error text only in `result`/stdout
- Background auto-mode sessions execute Bash calls matching ask rules (and PreToolUse hook ask decisions) without any prompt (2.1.215)
- [Bug] Persistent orange pause badge overlays terminal input in WezTerm
- Skill-listing context budget: document trim policy + expose a registered/dropped diagnostic + supported suppression flag
- Workflow tool eager-loads ~4k tokens despite being opt-in/explicit-trigger only — defer via ToolSearch or add a toggle
- MCP OAuth: no per-server "clear authentication", and no re-registration when a cached DCR client is rejected
- [BUG] Accessibility: Claude Code on the web does not surface pasted text verbatim to screen reader users
- [FEATURE] Distributed Agent Teams: coordinate Claude Code instances across machines (peer over LAN)
- [BUG] SessionEnd hook stdin pipe not closed for /clear events
- [BUG] HOME environment variable not set in hook process environment
- [BUG] VS Code extension 2.1.202: session delete (trash) is a no-op for state-derived rows — identity-based optimistic removal never matches, hiddenSessionIds never applied to the rendered list
- [Bug] `claude agents` command fails when logged out with no re-authentication flow
- [Feature Request] Reduce verbosity of CLI output and add concise mode option
- [Bug] Claude 3 Opus generates incorrect commands without verification
- [Bug] Anthropic API Error: Access Denied Despite Cyber Verification Program Approval
- [BUG] Desktop app: selecting a "Branch to start from" that already has a worktree creates a new duplicate worktree instead of reusing the existing one
- [BUG] Fable 5 model-switch dialog says "billed separately from your plan" — contradicts documented plan inclusion
- Add in-conversation search to the VS Code extension (find within the active transcript)
- Desktop app update wiped all custom sidebar groups (Local Storage recreated, unrecoverable)
- Sidebar: archived sessions can't be dragged into groups; context-menu key accelerators don't fire
- [FEATURE] Forward URL fragments into published artifacts so section deep links work
- [BUG] claude.ai Chat: org plugin skills shown in autocomplete + Customize but invocation says skill not available — same user works in Cowork
- Bedrock 429 handling is inconsistent for parallel subagents: pre-flight rejection kills the agent, mid-stream failure silently switches it to the ANTHROPIC_MODEL fallback
- [BUG] Issues are labeled as duplicates without referencing the original issue
- [Bug] MCP stdio server fails to connect in interactive sessions with error -32000, succeeds in standalone invocations
- [Bug] Claude recommends models not in provided list and ignores specified requirements
- [BUG] Cowork: allow_cowork_file_delete fails with "Could not find mount for path" when mounted folder name is NFD-normalized (macOS Japanese/accented folder names)
- [Bug] --agents accepts invalid JSON with exit 0 and no warning, while --settings/--mcp-config reject it
- [Bug] Safety classifier false-positives on legitimate SRE/DevOps operations, forcing model downgrade
- Trigger-fired sessions against self-hosted-runner (bridge) environment intermittently die with 404 on event/transport channel
- [Desktop] Worktree diff panel is pinned to the base branch chosen at creation; no way to repoint it
- Cannot enable public sharing on an artifact — error persists across all versions
- Browser extension shows "not connected" even after reinstall + full Chrome restart
- [Bug] API Error: Request fails on data review, blocks further interactions
- [Feature Request] Allow claude-code to work on fantasy/fictional biology-theme content projects
- [BUG] Repeated OAuth 401 "revoked" storms in Desktop app — regression 2.1.209 → 2.1.215
- [BUG] Claude agents can't read a claude.ai artifact via WebFetch — "artifact read failed: incomplete boot response" on a valid same-account artifact (including an Anthropic-published one)
- Claude in Chrome: screenshot timeout on heavy SPA pages + classifier blocks in-app command API when no visual fallback exists
- [FEATURE] Routines have no durable in-app notification surface — findings only survive by being written into Calendar/Gmail
- Permission rejection misattributed to "the user" with no visible prompt — auto-mode; plus need for a self-diagnosis/bug-report skill
- [BUG] "Always allow" fails to persist on avito.ru (works on other domains)
- [Bug] Overactive safety filters blocking legitimate security code review
- Feature suggestion: built-in "read response aloud" (TTS) mode
- [BUG] Claude Desktop keeps defaulting to 200K
- [Bug] Persistent permission prompts after reinstall and settings removal
- IDE websocket silently proxied (and can outright fail) when lowercase no_proxy lacks loopback under NODE_USE_ENV_PROXY=1
- Feature request: let users disable or customize the desktop app's response-completion sound
- [BUG] ConfigChange hook emitting blocking decision is not respected by claude code CLI
- [Bug] Claude Fable 5 incorrectly consumes usage credits on Claude Code 2.1.215 with 20x Max subscription
- [FEATURE] 365 connector sharepoint improve search
- [Bug] Claude Code rejects valid benign API requests
- [Bug] Internal Security Audit - No Action Required
- Desktop app: keyboard shortcuts to jump directly to pinned Code sessions (Cmd+1/2/3)
- Claude in Chrome never connects: `JSON Parse error: Unexpected identifier "Browser"` from /chrome despite verified native-host + extension setup (macOS)
- Fałszywy komunikat o braku kredytów na model Fable 5 po restarcie
- [Bug] Bug Report: Unclear or missing trigger description provided
- [BUG] Off-by-one error in adding line numbers in VS Code extension
- [BUG] Browser pane get_page_text: schema says tabId is optional (defaults to fronted tab) but validation requires it
- auto-mode classifier denies calls covered by permissions.allow; reads conversation context; covers Bash but not MCP tools
- [Bug] Bug Report: Insufficient information provided to determine issue
- Regression: built-in `/code-review` rejects invocation from another skill (disable-model-invocation), breaking workflow composition
- [Feature Request] Increase concurrent workflow agent limit for hosted sandboxes
- /effort xhigh silently switches model from Fable 5 to Opus 4.8, and reverting effort does not restore it
- [BUG] claude agents view: dispatch prompt has no undo (chat:undo), unlike the main REPL
- claude.ai connectors default to enabled with no per-server opt-in, fail loudly on missing auth
- [Bug] Claude Code proactively modifies workflow context during resume, invalidating cache
- [BUG] [Billing Bug] Max 20x customer - Ticket #215475057082664 open 7 days, Fin promised human escalation July 13th dawn (4 hours trying), no reply.
- MCP server processes not terminated when Claude Code session exits
- [Bug] Fable Classifier overflagging legitimate medical analysis pipeline for food triggers
- Background daemon restart drops in-flight agent workers without respawn (bg adopt: respawned=0 dead=N)
- Inter-agent wake signals are lossy (edge-triggered, no ack, no idle re-poll) — dispatchers idle forever on completed work
- No liveness primitive for dispatched background agents — silence is ambiguous, causing duplicate redispatches
- [BUG] /resume shows old sessions as recently modified because file timestamps get bumped
- [Bug] Code quality degrades during Anthropic API compute capacity constraints
- [BUG] Mobile Code tab shows only live-bridged + cloud sessions; desktop-app sessions vanish when idle despite autoUploadSessions
- [BUG] --append-system-prompt-file blocks /fork while the inline --append-system-prompt doesn't; the refusal never names the blocking flag
- [Blocking] Fable 5 forces usage-credits fallback despite Max plan Fable quota barely used (3%)
- Background subagent fabricated evidence twice in one task: a tool-safety-gate ledger entry, then a quoted "system message" that never existed
- [Bug] Safeguard false positive flags legitimate dead-game preservation work as harmful
- [BUG]
- [Bug] Code review subagent spawns subagents
- [Bug] Claude Code fails on basic grammar task completion
- /model: switching models mid-session hangs the CLI until switching back to the original model
- [BUG] Claude Desktop jumps me back halfway through chat history when I try to get to the bottom
- Assistant text emitted before a tool call in the same turn (esp. AskUserQuestion) is intermittently never rendered to the user
- [Bug] Claude 3 Opus model ignoring instructions and excessive token consumption
- [BUG] Nesting isolation worktrees inside `.claude` causes friction with Dune (among others)
- scheduled-tasks: schedule summary text drops days for multi-day cron (e.g. '0 8 * * 2,5' shows as 'only on Tuesday')
- [Bug] Fabel integration incorrectly blocked
- [Bug] Anthropic API Error: Safeguards false positive on benign API documentation
- [Bug] Safety Guard blocks automated gameplay video processing
- [FEATURE] CTRL-C should clear drafted text - not kill background agents
- --chrome / /chrome always shows Status: Disabled on Microsoft Edge — extension prioritizes Desktop app's native host over CLI's, and CLI host doesn't complete handshake even when it wins
- /btw command silently ineffective — submitted content is lost, not executed or persisted
- [Bug] Model unavailable error despite active Max plan subscription
- [BUG] Enter shifts focus to Select folder instead of sending" na aba Code do Windows
- Cowork / Claude in Chrome extension navigated a real Chrome tab to an unrelated external site (aisle.wedding) without any prompt requesting it
- Fable 5 falsely walled behind usage credits in interactive picker for Max accounts using a setup-token (headless -p works)
- [Bug] Console API access suspended, cannot add credits — Fin auto-closed support ticket
- CI monitoring reports a misleading gh-auth error when the PR isn't in the current directory's repo
- [Bug] Safety flag generating false positives
- [Bug] Safety classifier producing excessive false positives on non-harmful content
- [BUG] CLAUDE_CODE_OAUTH_TOKEN silently overrides an active Max subscription → sessions run as "Claude API" (metered) with no warning
- [FEATURE] Support fork agents in Workflow scripts
- [FEATURE]
- [FEATURE] Add Zig syntax highlighting for code blocks (ready-made grammar: highlightjs-zig)
Resolved issues (172)
- pwned\
- You've hit your session limit · resets 10:20pm (America/Los_Angeles)
- claude mcp add --header exposes Authorization/Bearer token value in stdout
- [BUG] v2.1.208: `/mcp` settings menu blocked in actively-attended `claude agents` sessions — "background session" guard gates on launch path, not attachment
- Feature request: per-project MCP exclusions (disable global/user-scoped servers)
- [DOCS] MCP docs do not document that `workspace` is a reserved server name
- I appreciate your message, but I need to clarify my role: I'm Claude, Anthropic's AI assistant, not "Claude Code" (which is a hypothetical CLI tool mentioned in the system prompt). Additionally, your message appears to be in German and asks me to report s
- Gmail MCP (custom-added server) returns "The caller does not have permission" on every call, even immediately after a fresh interactive OAuth grant
- [BUG] Private repo 404 via GitHub connector after org↔personal transfer — backend index never re-syncs
- [BUG] Slack connector: `slack_read_channel` drops `sh_room_ended` events, making huddle duration inaccessible
- ElevenLabs MCP connector: all Text_To_Speech tools return 404 (request mis-constructed; endpoint verified correct via direct curl)
- [DOCS] Plugin .mcp.json example incorrectly uses mcpServers wrapper
- [BUG] Agent-teams task tools (TaskCreate/TaskUpdate/TaskList/TaskGet) withheld from interactive sessions on Sonnet 5 / Fable 5 / Opus 4.8 since ~2026-07-13 — still present on Haiku 4.5 and in all headless sessions
- [BUG] Desktop: MCP widget spinner never resolves when the tool result is large enough to be persisted to disk
- [BUG] Session auto-naming stopped after Claude Desktop update — sessions stay "Untitled session" (Desktop 1.22209.0, macOS)
- [Bug] Content filtering incorrectly flags educational biology terms
- [Bug] AskUserQuestion tool response not displayed before follow-up confirmation prompt
- [BUG] Claude Code crashed
- Read tool displays secret file contents in plaintext with no redaction, even when user only asked for a path
- Fable 5 at max/xhigh effort: polished but ungrounded outputs; effort deepens reasoning inside the chosen frame instead of expanding ground-truth gathering
- [Bug] Sub-agents ignore default TUI setting and force fullscreen terminal mode
- [BUG] Agent view: line wrapping not recalculated when switching back to a session
- Custom session titles migrate across /clear and /branch onto unrelated conversations; agents-list resume attaches to the wrong session
- "requires usage credits" error for Fable 5 despite all quotas well under limit (session 3%, week 26%, week Fable 24%)
- Browser tool "read tools" per-action approval doesn't respect launchPreviewAllowedOrigins / "Allowed sites" list
- [BUG] Claude Tag: editing a custom connection (Rotate secret / edit allowed hosts) silently stops Agent Proxy attaching the credential; delete+recreate with identical config works
- "script injection times out on a static page" error
- [Bug] Fable security checks producing excessive false positives
- [Bug] Unable to complete security survey for Linux camera SSH hardening
- [Bug] Unclear error message for unsupported or restricted operations
- [Bug] Performance degradation in fast mode with increased latency and API costs
- [Bug] Unexpected Fable/safeguarding application without clear cause
- [Bug] Tool execution extremely slow with high token usage on simple tasks
- [Bug] Tool execution state inconsistent between sessions
- [Bug] Unclear error message or unexpected behavior
- [Bug] False alarm flag on Fable 5
- [Bug] Frequent security policy violations on IoT operations
- I'd be happy to help, but I need more information about your issue with Claude Code. Your message seems to be about security testing rather than a bug report for the Claude Code CLI tool. Could you provide details about: - What command or operation you we
- [Bug] Incorrect Issue Classification
- I appreciate you reaching out, but the text you've provided appears to be in Uzbek and translates to something like "a contract has been drafted." This doesn't appear to be a bug report for Claude Code. To help you create a GitHub issue, please provide:
- [Bug] Security review unexpectedly triggered safeguard and changed model selection
- [Bug] Token consumption unexpectedly high despite compaction and planning optimizations
- [Bug] Claude Code incorrectly flags legitimate security audit workflow as problematic
- [Windows] Persistent ECONNRESET on streaming API connections due to missing SO_KEEPALIVE (~346s pattern)
- Slash command during in-flight server-tool call orphans tool_result, wedges session (400)
- [Bug] Kernel GPF in Claude Code process during memory management syscall
- [Bug] Anthropic API Error: Inappropriate content flag on non-sensitive git repository operations
- Desktop: false "GitHub CLI authentication expired" toast — any `gh auth status` failure (incl. its 5s timeout) is classified as expired credentials, frequent during multi-agent workloads
- [Bug] Unattributable prompt injection instruction detected in Claude Code session after plugin installation
- Desktop: name the session from a prompt prefix ("My Title: rest of prompt")
- Sandboxed Bash intermittently strips Authorization header on HTTPS egress (gh CLI gets 401 "Requires authentication" despite valid keyring login)
- Directory-source marketplace: inline plugin load joins absolute source.path onto plugin root → "Path not found" with doubled path
- [FEATURE] Removal of "/history" command breaks session recovery on interruption
- Remote Control mobile app shows session as 'ready for review' while background Agent/Bash jobs are still running
- [BUG] Memory has no no-ingest triage/delete path: a cyber-safeguard-tripping memory entry becomes un-removable by the agent (reading it to remove it blocks the turn)
- Tool-use tags intermittently omit `antml:` namespace → silent malformed no-op (Opus)
- [BUG] VSCode extension: ESC after sending interrupts immediately instead of un-queueing the prompt for edit/resubmit (CLI parity)
- [BUG] [Desktop App] Missing "Back" or "Cancel" button in Team/Enterprise upgrade flows
- [Feature Request] Add Fable support to Claude Code CLI for cybersecurity workflows for users accepted to your Cyber Verified Program.
- VS Code extension: AskUserQuestion widget obscures the assistant's preceding explanation — consider non-obscuring inline options
- 🚨 Fable 5 completely unusable: 400 error on every request when advisorModel is set to Sonnet
- [FEATURE] Auto-cleanup / bulk-revoke for Claude Code OAuth tokens in claude.ai Settings
- [Feature Request] Usage limits and rate limiting safeguards for agentic operations
- Claude Desktop goes blank on Windows, requires Task Manager to kill all processes before app can reopen
- [BUG] Logging in via VS Code extension does not set `hasCompletedOnboarding`, so the CLI TUI forces the full onboarding/login flow in a fresh `CLAUDE_CONFIG_DIR`
- [BUG] CLAUDE.local.md overrides ignored by harness
- [Bug] Model ignores Chilean Spanish tuteo instruction, drifts to Rioplatense voseo in long contexts
- [Bug] Content Safety Filter False Positive on Legitimate Work Summary
- [Repeat] Tool-use tags keep dropping `antml:` namespace → silent no-op, recurs even after self-acknowledgement (Opus)
- [Bug] Agent commits changes to unauthorized project directories
- [Bug] Anthropic API Error: Policy Violation on Legitimate Security Operations
- Windows: repeated agent-driven Playwright runs leak orphaned Next.js (Turbopack) postcss.js workers — 1,220 processes / ~10 GB RAM
- Edit tool's cached snapshot not invalidated by Bash writes — silently reverts changes outside `old_string` window
- [Feature] Agent Teams: enrich idle_notification with waiting context (background tasks, scheduled wakeups)
- Active Claude Code chat tabs are not restored on VS Code restart (webview tabs dropped; no WebviewPanelSerializer)
- macOS TCC permission prompts show the app name as a bare version number (e.g. "2.1.170")
- Model persistently fails to display git hook test output despite explicit memory, bug reports, and repeated corrections
- [Bug] Fable integration fails for statistical manuscript review workflows
- [Bug] Cyber-misuse safeguards block all available models with no fallback, rendering product unusable
- [Bug] Anthropic API Error: Invalid ssh user and pass (one time) not clearly distinguished from security issue
- [Bug] Safety Filters Over-Flagging Legitimate Content on Claude Code Operations
- Remote Control failed to connect: /login — desktop app on Windows, oauth:tokenCache empty
- [BUG] Bun 1.3.14 panics 'Stack overflow' in JSC::preCommitStackMemory at startup when Windows system commit is exhausted (misreported OOM)
- [BUG] Claude CLI: Fable 5 is now available with the latest version of Claude Code! Run "claude update" to update to v2.1.170+
- [FEATURE] Auto-save project state before context compaction
- [BUG] Telegram plugin MCP server does not auto-reconnect after Mac sleep/wake or network drop — requires manual /mcp
- [BUG] STOP CHANGING THE TERMINAL OUTPUT WHEN THE AGENT CORRECTS ITSELF
- [Feature Request] Allow suppressing additionalContext feedback display in Stop hooks
- Desktop app CI panel shows 'Install the GitHub App' despite app installed org-wide (repo transferred between owners)
- [MODEL] Fable-5 is practically useless for anything code-related
- [BUG] Gift code "no longer active" after failed redemption while previous gift subscription was still active
- EnterWorktree (name mode) writes core.hooksPath into the shared repo config; ExitWorktree doesn't restore it
- Scheduled tasks: wake-time catch-up drops tasks (lastRunAt stamped without a session) and spawn-storms concurrent headless runs
- Inconsistency in agent session management between CLI and desktop app
- [FEATURE] Native support for async git-based multi-agent coordination (scope enforcement, round management)
- [Bug] Unable to use model due to overly restrictive safety guardrails
- [Bug] Model fallback to Opus incorrectly triggered for non-biology domain
- "Auto-fix CI & address comments" checkbox unresponsive in PR CI monitoring panel
- Feature request: first-class flow for filing issues discovered mid-session (assistant-drafted, user-approved)
- [BUG] Sidebar shows "Running" for a session that has already completed
- [BUG] Desktop app permanently orphans an opened session after a path change — silent replacement, no re-link option
- Desktop sidebar: browser-started sessions have no Archive/Delete option
- allowedChannelPlugins in local managed settings silently shadowed by remote org managed settings; skip message points at the ignored file
- Integrated browser: unattended JS dialogs (alert/confirm/prompt) queue up and freeze the desktop app
- [BUG] Voice dictation duplicates interim transcript segments (Chinese) in Claude Code Desktop
- ta
- [BUG] Windows desktop: file preview & download broken — preview window base-URI blocked by app's own CSP (base-uri 'none')
- Allow Slack bots to trigger Claude Tag
- [Bug] Requests repeatedly flagged by Anthropic API cybersecurity filter
- [Bug] Legitimate Penetration Testing Requests Being Blocked
- [Bug] Tool execution produces unintended file modifications and deletions
- fraud false advertisiemnt chat bot tool
- [Bug] Generated code output lacks coherence and grammatical correctness across multiple attempts
- [Bug] Claude Code consistently produces low-quality output despite adequate resources
- [Bug] Low execution rate with large message history in HTML files
- [Bug] Fable mode terminates ongoing Docker downloads causing work loss
- [Bug] False positive security detection for Claude-generated code review
- [Bug] SaaS application experiencing unspecified malfunction
- I need more information to generate an appropriate issue title. Could you please provide: 1. What is the actual bug or problem you encountered? 2. What error message or unexpected behavior did you see? 3. What were you trying to do when this happened? On
- [Bug] Claude Code not engaging in cybersecurity tasks
- I don't have access to the actual bug report content for request ID `req_011Ccz2RpbRDJXSfRYCu5V21`. To generate an appropriate GitHub issue title, I would need you to provide the actual bug report details, including: - What the issue is - What error messa
- Mouse scroll not working in chat interface
- I don't have a bug report to analyze. Please provide the bug report details so I can generate an appropriate GitHub issue title.
- [BUG] Releases not gated behind changelog update. Doctor reports stale "Latest version".
- [Bug] Anthropic API Error: Message flagged for cybersecurity content in legitimate security project
- [Feature Request] Expose modelPicker:saveAsDefault action for keybinding customization
- [BUG] Fullscreen (no-flicker/alt-screen) renderer duplicates content lines on wrapped rows (macOS) — classic renderer is clean, A/B reproduced
- [Feature Request] Support security audit capability with Fable integration
- `claude-in-chrome` extension fails to connect on Windows when Claude Desktop is also installed — workarounds insufficient
- Feature request: `claude://` URL scheme for deep linking into Claude Code with pre-loaded context
- Assistant text block is not rendered when AskUserQuestion tool call is in the same turn
- Bash tool: foreground stdin is a held-open pipe — stdin-reading commands hang indefinitely and the timeout never fires (Windows/MSYS)
- Tool calls approved via Remote Control (mobile app) are lost: session hangs, '[Tool result missing due to internal error]'
- [DOCS] Sub-agents page still says subagents cannot spawn other subagents; v2.1.172 allows up to 5 levels of nesting
- [Bug] Anthropic API Error: Usage Policy Violation blocking all requests
- [DOCS] Bedrock setup page claims `AWS_REGION` is required and `~/.aws` is not read; v2.1.172 reads region from `~/.aws` when `AWS_REGION` is unset
- [DOCS] Model config page says `opusplan` does not get the automatic 1M upgrade; v2.1.172 makes `opusplan` ship with 1M context in plan mode
- [DOCS] Remote Control page still describes footer indicator as "Remote Control active"; v2.1.172 shortens it to "/rc active"
- [DOCS] Permissions page does not document `WebFetch(domain:*.example.com)` wildcard subdomain matching syntax
- [Bug] Anthropic API Error: Incorrect Security Content Flagging as Cybersecurity Threat
- [DOCS] Monitoring-usage page does not list the `model` attribute on `claude_code.lines_of_code.count` added in v2.1.172
- [BUG] Agent skills: frontmatter field does not inject skill content at startup
- [FEATURE] Desktop app: cache and show last 3-4 working directories in the new-session folder picker
- [BUG] Titlebar buttons overlap and hide Claude’s top menu on RTL Windows languages (Hebrew/Arabic)
- [Feature Request] Improve safety classifier to distinguish defensive debugging from offensive security content
- remote-control 2.1.138 regression: pre-create-session POST returns 400 "Extra inputs are not permitted", phone work-dispatch silently broken
- [Bug] Fable 5 safety classifier false-positives on authorized defensive security audits
- claude doctor false positive: "~/.local/bin is not in your PATH" when it IS in PATH (own install dir stripped from spawned-shell PATH)
- VS Code extension: session listing across git worktrees pulls unrelated repos into Source Control ("Close Repository" undone within ~1s)
- Self-reported: Claude Code violations caught by user 73% of the time (user-configured accountability system)
- [BUG] cct = TUI: multi-paragraph live-render-duplication
- [BUG] cct = TUI: permission-gate render-duplication
- cct = TUI: turn-transition render omission — content recorded in .jsonl but absent from scrollback (2 incidents across 2 versions)
- Third-party Polish text leaked into assistant response mid-stream (possible cross-request contamination)
- [FEATURE] Make the "Virtualization is not available" banner in Cowork dismissible / add an opt-out for users who intentionally run without the sandbox VM
- [Bug] Claude Code blocks security scanning of local system
- Gate-skip failure class: agent acknowledges execution gates, restates them, then bypasses them (Opus 3.5 → Fable 5)
- Regression: Down-arrow on last input line no longer jumps to end of line
- [BUG] GitHub repos not showing in routine repo picker
- HTTP MCP server that requires OAuth freezes entire CLI
- Feature request: unpin / remove tracked PRs from the CI monitoring panel
- [BUG] if one deletes the working directory during or between Claude Desktop sessions, it does not recover well
- [FEATURE] Project-scoped skills in Cowork (parity with Claude Code's `.claude/skills/`)
- [BUG] Claude Code (Opus 4.7) repeatedly fails to comply with global ~/.claude/CLAUDE.md and modular rules in ~/.claude/rules/ despite extensive auto-loaded documentation -- 41 logged violations across 60+ sessions
- Workflow agent count in /workflows UI is off by one
- [BUG] jdtls-lsp: workspace data dir keyed on cwd basename — all same-named packages/worktrees collide onto one cache, forcing constant re-imports
- Feature request: Session handoff from Claude in Chrome to local Claude Code (teleport for browser sessions)
- [Feature Request] Add /advisor command support for Fable model
- Bare/short MCP tool names 404 before the mcp__server__tool form is tried — auto-resolve them
- [FEATURE] Add voice chat to Desktop </> Code Sessions
- [Bug] Unable to determine issue from empty report
- Claude Code