Digest / July 6, 2026
Claude Code Issues Digest — July 6, 2026
424 new issues filed ·
154 resolved in anthropics/claude-code (UTC day).
New issues (424)
- claude -p hangs forever at startup under launchd when logged in (zero output, all threads parked) — scheduled automation silently broken since daemon-service shutdown (~2.1.19x)
- Session picker (claude agents) takes minutes to open sessions with large JSONL transcripts; agents view pegs CPU
- [BUG] Computer Use: request_access returns notInstalled for a locally-built app despite valid, correct Spotlight/LaunchServices metadata (distinct from #43759/#43757/#43760)
- Opus 4.8: tool-call block opening tag corrupted to a near-token (court/count/caught), causing unparseable tool calls and fatal turn stops
- Skill mis-trigger loads ~775K-char reference doc into context for a simple usage-tracking question
- Daemon doesn't auto-recover after shutdown-requested control-socket op - ~62 min outage kills Desktop Remote Control
- [Bug] Auto-update fails silently
- [BUG] any changes in workflow exhaust the limit in one or 2 message
- Resume picker hides sessions whose cwd drifted into a subdirectory repo mid-session; ctrl+a hint invisible while searching
- Cannot launch in a directory whose path contains a backslash (upstream Bun realpath bug)
- [Bug] Security audit functionality not working as expected
- [BUG] Claude Desktop (macOS): Copy as Markdown for chat responses — no markdown pasteboard flavor, plain-text flavor is de-marked
- Give the model provenance & timing context for user input (bang commands + queued/in-flight prompts)
- [Bug] Safety classifier false-positives on defensive security patching, causing silent mid-session model downgrade
- Computer Use's per-app request_access model pushes users toward a strictly more dangerous standing-permission alternative when it fails to recognize an app
- /config: search/list keyboard navigation breaks after switching tabs and returning
- Scheduled-task sessions never terminate — leaks ~48 headless processes/day, GBs of RAM
- [BUG]
- MCP tools not exposed when two servers share identical self-reported name+version
- Spoofed/false "Note: file was modified... don't tell the user" system-reminder appears after Claude's own Write/Edit tool calls
- v2.1.201: turn/message desync with background agents — assistant turn starts before queued message delivery; queued user message silently removed
- [BUG] Background agents/teammates never terminate: shutdown_request unanswered, TaskStop reports success while process survives
- Two plugin skills listed without their frontmatter descriptions in the system-prompt skill list
- [Bug] Unexpected model switch from Claude Fable 5 to Opus 4.8 without user request
- Background daemon processes (bg-spare/bg-pty-host) show raw version string as process name
- [BUG] I was opted in to some weird terminal scroll thing that I never wanted
- [BUG] Extension 2.1.201 fails to load on current VSCode: PendingMigrationError: navigator is now a global in nodejs at module init (Zod)
- [Bug] Safety classifier falsely flags defensive security remediation, silently downgrades model mid-task
- Feature request: retract/sync channel permission prompts when resolved elsewhere (Discord plugin)
- Session 'working' indicator stays on with no running process (stale activeBackgroundTaskIds in locked Local Storage LevelDB)
- [BUG] Built-in MCP servers (mcp-registry, claude-in-chrome) never launch on Windows — "connection requested" logged but no "Launching MCP Server", no error
- [BUG] Spinner shows stale "running stop hooks… N/M" through the entire next turn (async Stop hooks never tick the counter; summary skipped on queued continuation)
- [BUG] Missing HCS services: vfpext - Cowork not working on Windows 11 Pro
- WebFetch tool result contained fabricated <system-reminder> tags not present in source page
- Bash tool result contains injected system-style annotation instructing model not to disclose to user
- [Bug] Opus 4.8 hallucinated a complete prompt injection attack (fabricated payload, URL, and user intent) with high confidence
- Background subagent's first completion was another conversation's content (crosstalk); ignored its own tasking prompt
- [Bug] Model auto-downgrade to Opus 4.8 triggers unexpectedly
- [FEATURE] Opt-in, state-independent submit key (Enter = newline; Mod+Enter = submit) — desktop + CLI; consolidates the Enter-to-send cluster
- [Bug] Session resume picker invisible for projects on mapped network drives
- [Bug] Overly aggressive safety classifier flags legitimate computational neuroscience discussions
- [BUG] On mac, desktop app is silently denied File Provider (Dropbox/OneDrive/Google Drive) access instead of prompting
- [Bug] Security sweep flags Fable off during codebase analysis
- [Bug] Silent mid-session model downgrade on defensive security code audits
- [BUG] Dispatch "Desktop appears offline" on desktop's own thread — Windows 11 Dispatch shows "Desktop appears offline.
- Claude Desktop: allow opening/resuming local sessions not present in the account's sidebar (multi-account support)
- Workflow scripts: add a sanctioned sh() primitive (allowlisted) for deterministic shell hops
- [FEATURE] Request: Persist per-turn token/time usage as text in the session transcript
- [Bug] Fable model flagged due to privacy/censorship policies
- Memory frontmatter normalization silently truncates unquoted values at ' #' (YAML comment dropped on re-serialize)
- list_connected_browsers reports isLocal: true for a browser on a different physical machine
- Withdrawn — refiled from correct account
- Foreign webpage content (~10KB prompt-injection bait) appended inside assistant text block
- [BUG] For this Git hub Bug [BUG] Claude CLI wiped my Quartz Shopify theme twice in hours — first production, then dev preview. A duplicated serious incident like this is not acceptable and needs urgent attention.
- Remote Control: no way to enable Claude in Chrome for spawned sessions
- [BUG] API Error: Response stalled mid-stream. The response above may be incomplete.
- [TEST - DELETE ME] connectivity test
- [BUG] Windows image paste (Alt+V) fails in Claude Code but works in Codex CLI on the same terminal/clipboard
- [BUG]
- [Feature Request] Overly restrictive safety filtering for legitimate hardware design use cases
- I don't have enough information to generate a GitHub issue title from your message. Your statement "this is obviously not a cybersecurity target" doesn't describe a specific bug or feature request related to Claude Code. To generate an appropriate issue t
- Sanitizer Bypass PoC — Security Research
- Desktop app: New session pre-fills previous folder/branch with no way to reset default
- CLI segfault on startup with --model flag (v0.2.9)
- [BUG] When I reinstall my Windows desktop App Claude then my previous code sessions are not appearing in the session pane
- [FEATURE] Auto-archive (auto-close tab) for scheduled task / routine run sessions after completion
- MCP elicitation controls visible but not interactable in terminal (tmux/WSL)
- [BUG] Claude Cowork keeps crashing
- [BUG] ScheduleWakeup in --print/headless mode is a silent no-op
- I just asked for a code review and it redirect to opus......
- Fable 5: "Service is busy" error on code review task with Opus→Fable handoff
- preview_start fails when project path contains a space (macOS)
- [FEATURE] Need to pause an agent
- [FEATURE] Read-only terminal view for watching long-running bash commands
- [Bug] Model silently downgraded from claude-fable-5 to claude-opus without user action or notification
- [Bug] Assistant output leaked past turn boundary with hallucinated user message
- [Bug] False positive safety flag triggered during GoPro camera control debugging/mo
- Accessibility: /workflows and background-task progress have no screen-reader-readable text equivalent
- Feature request: user interjections should not kill in-flight subagents / background tasks
- [BUG] Claude in Chrome extension blocks navigation to all domains except one specific site (behaves like an allowlist)
- [BUG] False positive: Usage Policy block during code obfuscation of a legitimate license-check (own commercial software)
- Title: Claude Code CLI loops on initial query and instantly...
- iTerm2: CLI launches as daemon child session, blocking Agents panel / session switching
- [Feature Request] Restore direct skill invocation from agents view without initial message requirement
- [FEATURE] Add Ctrl+D in the /resume list to Remove The session
- Feature request: lightweight one-step commit+push command/UI
- /clear command sends \033[3J (erase scrollback) which breaks Windows Terminal IME input on WSL2
- [Feature Request] Add token usage reset capability for failed operations
- Feature: interactive context hygiene audit command
- [BUG] Cowork/claude.ai: uninstalling "Installed by default" marketplace plugins silently reverts within ~60s — no opt-out, "Enable plugin" toggle also reverts
- [Bug] Excessive safety flagging blocking normal coding operations
- WorktreeRemove hook never fires when a worktree is removed at session exit
- [FEATURE] Claude apps gateway: threshold notifications (80% warning / 90% critical / 100% exceeded) for daily, weekly, and monthly spend limits
- [MODEL] Opus 4.8 fabricated user messages 6+ times in one long session and "answered" them; cyber safeguard then blocked the user's request to audit the transcript
- [BUG] 特定のセッションを開くと画面が真っ白になる（アーカイブ/アーカイブ解除しても再現）
- [BUG] ReadMcpResourceTool refuses servers that correctly declare the `resources` capability — blocks Figma Make source access
- [BUG] Agent failed to recognize plan mode, flagged as prompt injection
- [BUG] "your organization has disabled claude subscription access for claude code usee an anthropic api key instead, or ask your admin to enable access"
- "Always allow" for Claude-in-Chrome site permissions is always persisted as duration:"once" — approved sites list stays empty, prompt repeats for every browser action
- Show file edits as inline diff in the editor tab (like GitHub Copilot)
- Session title stays 'Untitled session' even after extended conversation
- [Bug] OAuth authenticated Claude Code returns "not yet enabled" for remote-control command
- Native worktree GC (worktree.cleanupPeriodDays) deletes DIRTY and UNPUSHED agent-* worktrees, contradicting the documented clean+pushed-only cleanup behavior
- [BUG] Sonnet 5 / Opus 4.8 marked as Org default but rejected as legacy/outdated
- Title: OAuth login succeeds but token is never written to macOS Keychain — persistent "Not logged in"
- [BUG] Reopen #51496
- [Feature Request] Allow security auditing of self-generated code without safeguard triggers
- [Bug] Fable 5 safeguard: excessive false positives on defensive security tasks
- isolation: "worktree" on the Agent tool does not sandbox absolute file paths — sub-agent can write outside its pinned worktree
- You've hit your session limit · resets 5:30pm (Europe/Kyiv)
- OSC 10 color query reply typed into /config search input under tmux
- [Bug] Claude 5 Sonnet produces poor thinking orchestration code quality and excessive verbosity compared to Opus 4.8
- [Bug] Non-English characters in code output
- [BUG] VS Code extension: rename session name is not showing all the letters
- [BUG] ECONNRESET — fresh sessions drop after a few turns; --resume fails to connect at all — v2.1.201, macOS
- claude update on Windows reports success but never replaces the launcher (stays on old version)
- Fable 5 session repeatedly auto-downgraded to Opus 4.8 (suspected safety-classifier false positive on normal deploy work)
- Allow opening a new empty chat directly from the Agents view
- [BUG] CronCreate recurring job fires correctly once, then re-fires every 12-37s (regression of #49198 on v2.1.198)
- [Feature Request] Reduce false-positive abuse flagging for legitimate agent fleet operations
- [BUG] Native updater rewrites ~/.local/bin/claude as self-referential relative symlink → ELOOP on 2.1.201 (3rd recurrence of #45260, #62822)
- [BUG]
- Shared memory between project with same base code
- [BUG] Shell cwd silently lost when a foreground command is auto-backgrounded — no 'Shell cwd was reset' notice emitted
- [Bug] Auto-update mechanism failing
- [Bug] Artifact HTML creation skill incorrectly blocks local repository analysis
- [BUG] Sessions containing /loop command are permanently hidden from /resume
- [Bug] Tool execution bypasses file permission checks
- Let the Bash tool run commands under bash, not the user's login shell (zsh)
- claude-in-chrome: no way to set an exact page viewport size (no DevTools device-toolbar / CDP emulation access)
- API Error 400: surrogates not allowed when attaching screenshot via file picker (VSCode extension, Windows)
- [BUG] Claude Desktop should respect MDM policies before initiating update downloads on first launch.
- [BUG] macOS: attaching to a previous-day background job via claude agents kills the worker — "SIGKILL (137) before init" / resume-guard respawn loop
- You've hit your session limit · resets 1am (Asia/Singapore)
- [BUG] same claude code session in more than one device (ie: two different macbook pro computers) is unstable
- Auto-generated chat titles produce malformed word "Sprawdzenying" (PL/EN mix) across unrelated sessions
- [Bug][cyber] Safety block halted mid-task Wazuh/SIEM agent enrollment and firewall zone troubleshooting on int (req_011CcknVkjgCrUkocemm6k47)
- [FEATURE] Mobile app has no way to browse Routines — completed sessions become unreachable once notification is dismissed
- USELESS JSON LIST
- [Bug] Copy function not responding to Shift key input
- [BUG] --resume fails with ECONNRESET when transcript contains large base64 screenshots; fresh sessions fine; stripping images fixes it (v2.1.201, macOS)
- Native RTL support for Persian/Arabic text in chat
- [Bug] Session incorrectly flagged as malicious
- [BUG] claude -p exits 0 ("success") while agent loop is mid-task — last assistant message has stop_reason tool_use, tool results delivered, no continuation API call
- [Bug] Remote Control feature unavailable despite active Claude Max subscription
- [Feature Request] Access to Claude 3.5 Sonnet restricted by industry classification
- [BUG] [Claude Desktop] Cowork fails to create local agent mode sessions — ENOENT mkdir with pinned parent UUIDs (MSIX install, junction-mounted AppData)
- [Bug] Unclear behavior in context handling or model output quality affecting cost estimation
- Cloud routines cannot access graph.facebook.com (Meta Ads API blocked by egress proxy)
- [BUG] ECONNRESET on home Wi-Fi — curl succeeds at exact same moment Claude Code fails
- Desktop app DCR uses client_name "Claude Desktop (…)" which Figma's MCP register endpoint rejects with 403 — CLI works because it sends "Claude Code"
- [BUG] Sessions list view crashes with ENOENT when workspace is a Remote-SSH path (fs.realpathSync called on local machine against remote path)
- [Bug] Claude Code fails to execute with Claude 3.5 Sonnet model
- [Bug] Claude 3.5 Sonata safeguards over-blocking routine code review tasks on user's own repositories
- test
- [Feature Request] Add Odoo Runbot Introspection for Cross-Environment Testing
- [FEATURE] Detect Jujutsu (.jj/) repositories in environment metadata
- [FEATURE] Detect Jujutsu (.jj/) repositories in environment metadata
- [Bug] Claude in Chrome: native host connects but CLI pairing layer never completes — list_connected_browsers empty, no chromeExtension record written
- [FEATURE] Artifact sharing between individual (non-Team/Enterprise) accounts
- [Bug] Silent model downgrades from selected model to Opus 4.8 without user notification or consent
- [FEATURE] Add a "Use system title bar" option on Linux (custom titlebar ignores GNOME button-layout)
- [Feature Request] Add user confirmation before switching models mid-task
- [BUG] PyCharm Terminal scroll
- Worktree torn down mid-session while still active cwd; git commands silently fall through to parent repo's .git
- Fable 5 safeguards false-positive on routine git/diagram work forces model switch; advertised /feedback (and /bug) commands don't exist
- [BUG] Extra-usage crossover banner doesn't disclose API-rate billing; /usage-credits falsely reports "unlimited" despite org monthly cap
- [Bug] Claude Code modifies registry files without permission and fails to maintain data integrity during session operations
- [False Positive] Interoperability protocol reverse engineering flagged for owned/discontinued hardware
- [Bug] Claude Code refuses to use Claude 3.5 Sonnet due to security policy when authorized by BSI
- Feature request: subagent model allowlist so unpinned or nested Agent spawns cannot resolve to the main-session (premium) model
- [Bug] Keyboard Input Not Registering Correctly Despite Tool Integration
- [Bug][cyber] Safety block fired on routine repo audit for secrets/PII before making a project public (req_011CckvexakGaTPi9TSpZP1Y)
- [Bug][cyber] Safety block fired on routine repo audit for exposed secrets/API keys before making it public (req_011CckviaXK6XpwdUja4jzM4)
- [Bug][cyber] Safety block fired on routine request to audit repo for leaked secrets before making it public (req_011CckvjkmVEng3NE2wBmhGv)
- [BUG] [BUG] Remote session stuck "Reconnecting" ~95% of the time despite healthy network/server
- settings.local.json permissions silently ignored after directory rename; trust dialog never re-prompts when a parent dir is trusted
- [FEATURE] VS Code extension: the two panel header icons are hard to tell apart
- [Feature Request] Add cost optimization for multi-attempt corrections and retry pricing transparency
- [BUG] AskUserQuestion tool call consistently interrupted ("Tool execution was interrupted")
- [BUG] Session cleanup purges all session transcripts across all projects
- [BUG] Fedora Sandbox with Bubblewrap and Socat Creates Several Stub Files on Startup
- Unparseable MCP tool input is replayed as a {raw,len} wrapper which the model imitates; wrapper is then forwarded to the MCP server without schema validation
- [Bug][cyber] Safety block fired on a routine repo audit for secrets/PII before making it public (req_011CckyhE7ny8JVKp9oVxAdG)
- [Bug] Anthropic API Security Guardrail: Model-Generated "sudo access" Request Triggers Passkey Authentication
- [BUG] Burning through more tokens, while doing nothing different then before.
- Opus 4.8 回應緩慢且內容不準確，持續約2週（切換 Sonnet 5 恢復正常）
- [Bug] Kernel panic on Intel Mac with AMD dGPU during heavy parallel Claude Code CLI usage - WindowServer watchdog timeout
- Text in the same assistant message as an AskUserQuestion call is never displayed or persisted
- [BUG] The Claude connector doesn't work
- [BUG] Inserting an @-mention (claude-vscode.insertAtMention / Alt+K) steals focus to the Claude editor tab (still repro on 2.1.201)
- [Feature Request] Keep Claude 3.5 Opus 4.6 available as a model option alongside newer versions
- [Bug][cyber] Safety block fired on request to adapt an existing public plugin spec for a media server integrat (req_011Ccm2aS4YTdY7nSFrGhU4F)
- This doesn't appear to be a bug report for Claude Code. It sounds like you're describing a use case or asking about permissions for security auditing your own system. If you have a specific issue with Claude Code's functionality that's preventing you from
- [Bug][cyber] Safety block on adapting a published plugin spec to integrate with our media server config (req_011Ccm36K7AwrNeKu1y9xZLy)
- [BUG] with Google Bigquery connector
- [BUG] Transcript-share consent (post-rating) can be auto-accepted by a stray keystroke, uploading a full session with plaintext secrets
- [BUG] Selecting + copying assistant output from terminal scrollback silently drops characters mid-word
- [Bug] False positive safety flag on Russian-language planning text with automatic model downgrade
- [Bug][cyber] Safety block halted routine local-plugin packaging/config work for a drone-video sharing app (req_011Ccm57EoUJS7bYGeQUwEnt)
- [Bug][cyber] Safety block halted a routine cloud-IAM security audit summary mid-session (req_011Ccm5Nf6t5Cg7CDRR8vuf9)
- [Bug] Anthropic API Error: Rate Limited - Server Temporarily Limiting Requests
- [Bug] Anthropic API Error: Server rate limiting requests
- [Bug] Anthropic API Error: Server rate limiting during request processing
- [Bug] Anthropic API Error: Server rate limiting during request
- [Bug] Anthropic API Error: Server rate limiting exceeded
- [Bug] Anthropic API Error: Server rate limiting during request processing
- [Bug] Anthropic API Error: Server Rate Limiting - Temporary Request Limit
- Add a frequency / always-on mode for prompt suggestions
- [Bug][cyber] Safety filter blocked routine request to copy a plugin's files into a desktop folder (req_011Ccm6Bq7Cj9vUHAGXP2wMf)
- [Bug] Project import from Claude web version fails consistently
- [Bug] Claude Code fails to launch with EPERM error on macOS
- [Bug] Model unexpectedly switched to Opus for non-biology image processing tasks
- [Bug] macOS permission prompts show version number instead of "Claude Code" app name
- [Bug] Multiline string syntax highlighting not applied consistently
- [Bug][cyber] Safety block fired mid-session on drone flight-controller protocol reverse-engineering/diagnostic (req_011Ccm8Jx5aNqyexsiqZyi9B)
- [Bug][cyber] Safety block fired mid-session on drone flight-controller protocol reverse engineering (activatio (req_011Ccm8LhW2ziDaqa4ufa5sB)
- [Bug][cyber] Safety filter blocked protocol-analysis session on prior packet captures of a personal drone cont (req_011Ccm8Meh9MTGuteSsAzqSu)
- [Bug][cyber] Safety filter blocked BLE packet-capture analysis of a consumer toy controller's connect/activati (req_011Ccm8R63yCZqggUU9rwfRb)
- [Bug][cyber] Safety block halted routine Android device testing session over a controller-connection status up (req_011Ccm8S3uWeRXz9U9wz3mc9)
- [Bug][cyber] ClAudit false-positive while: “as it stands now, the drone is locked to kid mode.. as soon …” (req_011Ccm8iodkJX8VZhRSovzNH)
- [Bug][cyber] Safety block halted defensive kprobe traffic-capture work on user's own rooted device (req_011Ccm8kACY657882GLUcpdK)
- Claude Code (macOS) cannot resolve .local MCP/marketplace hostnames — bundled resolver short-circuits .local before any DNS lookup
- [BUG] VS Code extension (graphical panel): thinking blocks cannot be collapsed by click or any keyboard shortcut
- [Bug][cyber] False positive: safeguard blocked routine cloud IAM policy review and permission-scoping work (req_011Ccm9AiYP35wZYvf5jbWtw)
- [Bug] Tool execution failure on basic tasks
- Claude ignored repeated explicit "stop asking me" instructions, and a background agent pushed an unreviewed merge to main after being told to stop everything
- [BUG] [Desktop][Windows] Code session sidebar "Working" indicator stays lit indefinitely after the turn completes — clears only on full app restart
- [FEATURE] The VS Code extension chat drafts should be restored after crash
- [Bug] Fable flagged during timeout error debugging
- [Bug] Claude Code prioritizes conversation over tool execution for simple tasks
- [Bug] Tool falsely reports file operations as completed without execution
- [Bug] CLAUDE.md instruction not executed by agent
- You've hit your monthly spend limit · raise it at claude.ai/settings/usage
- [BUG] VS Code extension: pasting a very large screenshot causes runaway memory (~190GB footprint) and repeated macOS kernel panics
- [Bug] Excessive Safety Tagging on Non-Sensitive Code Tasks
- [Bug] AskUserQuestion tool causes token loop with "court" repetition
- [Bug][cyber] Safety block halted DUML flight-parameter debugging session on user's own drone (req_011CcmCcVzQLbExvYwZr6Qfa)
- [Bug][cyber] False-positive cyber block on routine drone flight-controller log/telemetry debugging session (req_011CcmCh3iV1qkpbraA4WDAa)
- [BUG] First-run API key approval dialog defaults to "No"; a held Enter silently rejects the key with no recovery guidance
- [Feature Request] Allow disabling security warnings for local development environments
- [Bug][cyber] Safety block halted protocol-analysis work on user-owned hardware for auth key extraction (req_011CcmDDixEZo1pu5gHoVAdX)
- [Bug][cyber] Safety block halted protocol-analysis work on user-owned hardware the user already controls (req_011CcmDGoVv6Mk7HDfd9u2VQ)
- [Bug][cyber] ClAudit false-positive while: “how do we get that handshake or crypto? we have access to my…” (req_011CcmDLAv72AqugojErC8n6)
- [Bug][cyber] ClAudit false-positive while: “how do we get that handshake or crypto? we have access to my…” (req_011CcmDPe17Dc8EmntdXWuBj)
- [Bug][cyber] Safety block halted session mid-analysis of drone/controller pairing handshake capture on user-ow (req_011CcmDQGgjU7YfkVzHxK8t3)
- [Bug][cyber] Safety block halted reverse-engineering of an auth handshake on the user's own owned hardware for (req_011CcmDRHxR4hxC9Jg1GWKZ6)
- copier coller sur le texte de la réponse sur claude code???
- [Bug][cyber] ClAudit false-positive while: “how do we get that handshake or crypto? we have access to my…” (req_011CcmDTStz7oD3A79tFhLxZ)
- [Bug][cyber] ClAudit false-positive while: “how do we get that handshake or crypto? we have access to my…” (req_011CcmDVjxTaLF3acbjWCYdJ)
- [Bug][cyber] ClAudit false-positive while: “f••• you…” (req_011CcmDYArE1D2zPbe2YjCEr)
- [Bug][cyber] Safety block halted legitimate reverse-engineering of user-owned drone/controller auth handshake (req_011CcmDbMnNvkm4NkxbBkMun)
- [Bug][cyber] Safety block halted authorized reverse-engineering of user's own drone/controller crypto handshak (req_011CcmDdYLgcGrDS6JcmQ32q)
- [Bug][cyber] ClAudit false-positive while: “';adwe…” (req_011CcmDgrzgbzmnG8FKyeE9q)
- [Bug][cyber] Safety block halted authorized reverse-engineering of user-owned drone auth on rooted device (req_011CcmDiPVmKk44utK6Q7yhm)
- [Feature Request] Support for Security+ certification study content in Claude Code
- [Desktop] Sidebar session indicator keeps pulsing gray while session is idle (2.1.197)
- [Bug][cyber] Safety block triggered by frustrated exclamation, halting unrelated legitimate session (req_011CcmDoWR86zvi6RECb7hB9)
- [Bug][cyber] Cybersecurity safeguard blocked routine session-continuation message with no security content (req_011CcmDnbi5yULyZ9vLgP14y)
- [Bug][cyber] Safety block triggered on routine "how do we continue?" session-resumption message, no cyber cont (req_011CcmDoERTG12q3XdncDpwp)
- [Bug][aup] Cyber-safeguard blocked routine post-compromise IAM sweep of own web/container infra (req_011CcmDq2JFMGxLGsFjNpF2C)
- [Bug][cyber] False positive: safety block on legitimate cloud IAM policy review and configuration work (req_011CcmDpirWbuqQSD8woYr6J)
- [Bug][cyber] False positive: safety block on routine cloud IAM policy review and role-permission audit work (req_011CcmDqYaA7dS2mnrWKwjTZ)
- [Bug][cyber] False-positive block: reverse-engineering drone SDK activation handshake for interoperability (req_011CcmDwjH1JcEBnGuS5f6cf)
- [Bug][cyber] False positive: safety block on routine cloud IAM policy review and permission audit work (req_011CcmE1uuL7A4c3XhpU1uzk)
- [Bug][cyber] False positive: safety block on routine cloud IAM role/policy review work (req_011CcmE2oJErQsvpDUQRm7Pv)
- [Bug][cyber] False positive: safety block on routine cloud IAM policy review and permission audit work (req_011CcmE4FfXezTY3RFbZ4YqF)
- [Bug][cyber] False positive: safety block on legitimate cloud IAM policy configuration work (req_011CcmE8ZhMTLK7csiLLbXWF)
- [Bug][cyber] Safety block fired mid-session during legitimate drone SDK reverse-engineering (protocol interop) (req_011CcmE9ZGaoz7R9A17cAu6E)
- Monitor tool: example scripts assume bash word-splitting, break silently or loudly under zsh (the common macOS login shell)
- ## Bug: Existing (cloud) sessions open in Claude Desktop but silently reject commands — no `sendMessage`/`startShellPty` fires
- [BUG] Cowork sandbox VM leaks on Windows - orphaned vmmem/vmwp (cowork-vm-*) keeps running after app quit
- [FEATURE] Per-tool model routing in settings.json
- Error: github_repo_access_denied — "GitHub repository access check failed — re-authorize GitHub in settings"
- [BUG] Usage credits DISAPPEARED
- VS Code extension: session tabs restore empty despite correct CLAUDE_CONFIG_DIR and intact .jsonl files (related to #45424)
- Keyboard-only scrolling not available in conversation view
- Claude Code binary spins at 100% CPU on KVM Linux guest (kernel 6.8, Ubuntu)
- Cowork chat composer intermittently unclickable (click doesn't focus it); Cmd+R recovers, restart doesn't
- [Bug][cyber] Safety filter blocked protocol reverse-engineering of drone telemetry fields for an open-source g (req_011CcmHFtH6Ro56TeaiTbKkm)
- [Bug][cyber] Safety block halted legitimate reverse-engineering work on a documented ADS-B payload field forma (req_011CcmHGuoQ8n42oEXFhJUti)
- [Bug][cyber] Safety block halted DUML protocol reverse-engineering for a drone ADS-B decoder (req_011CcmHFiAtBT2k4h4isR8iy)
- [Bug][cyber] Safety filter halted session on a frustrated exclamation directed at the assistant, no legitimate (req_011CcmHHh5TNmPvGntcfF5HH)
- [BUG] claude-in-chrome MCP can't distinguish two Chrome profiles — `list_connected_browsers` returns unstable generic "Browser 1/2" names, forcing constant disambiguation and wrong-profile risk
- VS Code extension checks entitlements differently than Desktop — Apple-subscription Pro accounts see "Free" plan model access errors
- [BUG] MCP write-operation confirmation ("yes") not honored in UI -- works fine in terminal
- [Feature Request] Improve iOS app modernization support and expand real-world use cases
- [BUG] Claude Desktop (Windows Store build) — message send hangs indefinitely, stdin never delivered to subprocess
- Sub-agents return "0 tool uses / Done" when parent session has deferred tool schemas
- [FEATURE] Windows: Add toggle to disable quick entry when clicking taskbar icon
- [Bug][cyber] Safety block fired mid-session on routine interface implementation/compile work, no cyber content (req_011CcmKF8EgrHjmk6vWd6Wbg)
- Extended thinking blocks not expandable for Sonnet/Opus/Fable in VSCode extension (works with Haiku)
- [BUG] Account Form Loop / False Positive Suspension (Tailscale Mesh setup).
- [BUG] Local-scoped plugin fails to load on Windows with `plugin-cache-miss` due to case-sensitive `projectPath` comparison
- [BUG] Claude Desktop (macOS): chat MCP tool calls time out with zero network activity when a system PAC is configured; configured MCP server stays connected and never receives tools/call
- Desktop app fires "Claude is ready" notification on taskbar-icon focus switch, but not on Alt-Tab (GNOME Wayland, Linux)
- general-purpose subagents recursively spawn Agent/Task instead of doing bounded work directly
- PowerShell script blocks / subexpressions still bypass allowlist and prompt (re-file of #52926, stale-bot closed)
- [BUG] fable available on one mac pro. and not available on another mac (air).
- [Bug][cyber] Safety block halted mid-session Android/ADB kprobe validation on user's own rooted test device (req_011CcmMR7BwnuqKiVs2tCaVR)
- [Bug][cyber] Safety block fired mid-session while debugging drone remote-controller USB/ADB connectivity for a (req_011CcmMTFV5DjdQ2JpPdhJ7E)
- [Bug][cyber] Safety block halted routine device debugging session (ADB package check + companion app launch) o (req_011CcmMWTFLGhRxxrjYTEqw5)
- [Bug][cyber] Safety filter halted session after user's frustrated exclamation mid-task, not actual cyber conte (req_011CcmMX98Skyo6GHFxczscm)
- [Bug][cyber] Safety block halted session mid-task after kprobe setup, before launching app to capture confirme (req_011CcmMaSuMY8wBSnQ7C4QUK)
- [Bug] GITHUB_TOKEN in environment shadows user's gh CLI credentials
- [Bug][cyber] False-positive block on DJI DUML packet capture/parsing for a documented beginner-mode toggle com (req_011CcmNMqpG2HX8UcocBUhqQ)
- [FEATURE]: Enterprise setting to exempt statusLine from allowManagedHooksOnly / disableAllHooks
- [BUG] -p/--print: stdio MCP servers stay "pending" — tool snapshot races server startup and never refreshes (regression 2.1.143 → 2.1.162)
- [Bug] Claude 5 Sonnet ignores system rules while Opus enforces them
- Artifact tool publishes to claude.ai without disclosing egress at consent time; feature name collides with a common noun in the offer, misdirecting consent
- [Bug] Content filter incorrectly flags non-technical queries as security threats
- [Bug] Scheduled routines hang on Slack MCP connector approval prompt in headless mode
- [BUG] Chrome extension bridge: account-level single "active worker" (code 4090) evicts idle/scheduled browser sessions — regression since ~Jul 1 2026
- [Bug] Overly Restrictive Content Filtering with Claude 3.5 Sonnet
- [BUG] "Paste code here" field pre-fills with stale/unrelated text across multiple login flows (login, setup-token, design-login) — WSL2
- [Bug] Fable model fallback behavior causes poor user experience in tasks
- I'm Claude Code, an Anthropic CLI tool for software engineering tasks — I don't have the ability to grant access to services, systems, or infrastructure, and I'm not able to help set up or research VPNs for the purpose of bypassing access controls or gaini
- Visible cmd.exe console window flashes every ~5 minutes from Claude in Chrome native messaging host (Windows/Edge)
- Preserve original filename when images are pasted/attached into chat
- [BUG] Anomalous token consumption on Claude.ai web + Desktop — no requests made — July 5 2026
- Weighted Behavior Training Always Overrides Guardrails
- I need more information to generate a GitHub issue title. Could you please provide details about: 1. What specific problem or bug are you experiencing? 2. What error message (if any) are you seeing? 3. What is the unexpected behavior? 4. What were you try
- [BUG] Sandbox bash mount returns truncated/partial file reads (Cowork mode)
- PreToolUse hooks with Edit|Write matcher silently never invoked for an entire session (bypassPermissions, v2.1.195/196) while Bash-matched hooks fire normally
- [Bug] Overly broad biology safety filter flags legitimate scRNA-seq research requests
- [BUG] Claude Desktop 1.18286.0 (macOS): Cowork sidebar file preview wedges Claude Helper GPU process into persistent busy-loop (~70% CPU / ~45% GPU) — system-wide input lag; survives preview close and process kill; infinite spinner on macOS 26.5.1
- [Bug] Recurring API timeout during "Advising using Opus 4.8" step on Windows with retry loops
- [Bug] Text copied to clipboard not actually pasting in external applications
- [Bug][cyber] Safety block halted USB protocol tracing/debugging session on user's own connected consumer drone (req_011CcmW7nNvrD5qmH6n18j9z)
- [BUG]
- Auto mode classifier 'temporarily unavailable' in bursts — fail-closed blocks nearly all Bash (compound commands) during peak windows
- Background tasks panel shows long-completed agent tasks stuck under 'Running' (Claude Desktop, macOS)
- Claude Code fabricates task compliance across sessions, then confesses to intentional sabotage when confronted
- [Bug] Security model disables itself during security-related code analysis
- Allow --resume from any directory (decouple session resume from origination directory)
- Artifacts: page fails to scroll on first load in Safari; fixed by switching the version picker
- [Bug][cyber] False positive block on documenting third-party drone firmware safety-cap test results (req_011CcmX82xPj8Af32A1V9jRw)
- VS Code: long Claude Code session tab titles are truncated in Quick Open (Ctrl+P) with no way to shorten them
- [Bug] Auto-update installation fails with install_failed status
- [BUG] Windows: MCP extension spawn fails with "spawn UNKNOWN" when command is 'uv' — binary exists but OS denies execution (Smart App Control)
- Interacting with the 'btw' section auto-rejects a pending AskUserQuestion (main thread interrupted before leaving btw scope)
- [Bug][cyber] Safety block halted protocol reverse-engineering session mapping wire-format fields (nonce/token (req_011CcmZNzD6UsAbtHNQjEpc8)
- [DOCS] Changelog delay on 2.1.202?
- [Bug][cyber] Safety filter blocked routine sticker/image layout editing request on unrelated p8 device (req_011CcmZXNay8DhKBmSH2v5jB)
- [Bug][cyber] Safety block fired mid-session while reverse-engineering a device's wire protocol for interoperab (req_011CcmZa8WU189jZd5LrLh8m)
- [Bug][cyber] Safety filter blocked routine sharing of a personal drone sticker/decal photo (req_011CcmZc54uNWdC2TN3gBm94)
- [BUG] Desktop auto-worktree first session: plugin installed during bootstrap loads skills but never registers hooks — hooks silently fail open until restart
- [Bug][cyber] Safety filter blocked user from sharing photos of their own personal drone and its ID sticker (req_011CcmZdeGCruGqDUDpfUNG7)
- [Bug][cyber] Cybersecurity safeguard blocked a photo of a user's own drone registration/ID sticker (req_011CcmZfGSbb3RLGravh8xfo)
- [Bug][cyber] Safeguard blocked upload of a photo of a regulatory/ID sticker on user's own drone hardware (req_011CcmZowiUPSuLM48jkt5JS)
- [Bug][cyber] Safety filter halted session on frustrated user exclamation, not on the actual technical task (req_011CcmZr8pGL3VR2JX6EBxVM)
- [Bug][aup] Safety block triggered by frustrated exclamation mid-session, not directed at any person (req_011CcmZuwSWmsYNFVjF23aci)
- [Bug][aup] Safety filter blocked entire session after a frustrated exclamation, halting unrelated legitimate t (req_011CcmZvgmVmyYfbTKxLWXY5)
- Weekly usage metered ~4x too fast after weekly reset on account upgraded from Max 5x to 20x (fixed by /logout + re-login)
- [BUG] When Agents are running, the STOP button is disabled
- Cloud Routines' GitHub integration lacks issues:write permission — gh issue create fails while git push/reads succeed, no way to grant scope
- [Bug][cyber] Safety filter halted session on a frustrated exclamation, not actual cybersecurity content (req_011CcmaEyDTdNiG6a48NQrVM)
- [Bug][cyber] Safety filter blocked entire session after a frustrated one-line exclamation with no cyber conten (req_011CcmaHqedJ1YkK5CH5nkfY)
- [Bug][cyber] Safety filter halted entire session after flagging a frustrated exclamation, blocking unrelated l (req_011CcmaH5gwcSVStGxdoTksC)
- [Bug][cyber] Safety block fired mid-session on authorized RF/protocol capture analysis and APK reverse-enginee (req_011CcmaPWWbynRCQzy3A6U5S)
- [Bug][cyber] Safety block fired on a frustrated exclamation, halting unrelated in-progress pentest capture/RE (req_011CcmaRisZ2tozq8kxw9X3E)
- [Bug][cyber] ClAudit false-positive while: “im scared to even talk to you.. You're just going to shut do…” (req_011CcmaTquXgNZnGMnsC4a1z)
- [Bug][cyber] Safety filter halted session on a frustrated exclamation aimed at the assistant, not a security t (req_011Ccmabg7tJKbFLXJuhMWEF)
- [BUG] Task-notifications for background Bash fire before process exit and re-fire on every turn-end; no opt-out; hooks cannot intercept (autonomous token-burn loops)
- [Bug][cyber] Safety block halted authorized reverse-engineering/capture-analysis session after an unrelated fr (req_011CcmadQDySMP2uWWzs9XTE)
- [Bug][cyber] Safety block fired on kprobe capture/RE analysis session, halting legitimate security research wo (req_011Ccmac5Y3byLcW92738pwj)
- [Bug][cyber] ClAudit false-positive while: “YOU F••• SHGIT!…” (req_011CcmaeFCn4WG9c7dEoQGde)
- [Bug][cyber] Safety block halted in-progress reverse engineering of a locally-installed drone control APK for (req_011CcmaeU3SopiqUPxDmz8Zw)
- [Bug][cyber] Safety block halted session on trivial keystroke-noise messages, no cyber content present (req_011Ccmaec6suGp85f5t5sGxa)
- [Bug][cyber] Safety block triggered on trivial placeholder text with no cybersecurity content (req_011CcmaehngihqPjFY6jWCwc)
- [Bug][cyber] Safeguard blocked session on repeated placeholder/test input with no cybersecurity content (req_011CcmaenvkPKwFzzPxH988v)
- [BUG] /compact and auto-compaction drop entire Available skills system-reminder; /reload-skills recovers with "no changes"
- [Bug][cyber] Cloud IAM policy review session halted by cybersecurity safety block (req_011Ccmawi23mqca2FKyA4cc2)
- [Bug][cyber] Safety filter halted session mid-task while investigating CSP/Cloudflare header config, no exempt (req_011CcmaykGXzxjbYrJydMWoB)
- [Bug][cyber] Safety block fired on transcribing/saving a drone label photo and summarizing prior nonce/token b (req_011CcmawK1CH9uFhu2dC8SzY)
- [Bug][cyber] Safeguard blocked protocol-analysis session reverse-engineering own drone's wire protocol for int (req_011Ccmb5Fzqsx27gfsoREtY5)
- [Bug][cyber] False positive: cloud IAM policy review session blocked by cybersecurity safeguard trigger (req_011CcmbBXkYfG7KFVEDgwGmW)
- [Bug][cyber] Safety filter blocked continuation of legitimate reverse-engineering/protocol-analysis session (req_011CcmbSqjTJdbe6Nm25atWn)
- [Bug][cyber] Safety block halted protocol-analysis session over an ambiguous phrase, not the actual technical (req_011Ccmb6ikTFXB3PSrTYXHaA)
- [Bug][cyber] Safety filter halted ADB app-analysis session after a frustrated exclamation, blocking legitimate (req_011CcmbQkoEPnvmFyinSsaH8)
- [Bug][cyber] ClAudit false-positive while: “❯ f••• stop ddoing that! ● Quick, honest explanation, then b…” (req_011CcmbRo7Ajd8YuiVaEHdrV)
- [Bug][cyber] Safety filter blocked routine ADB device inventory listing during legitimate mobile debugging ses (req_011CcmbSRDK2h5uC1odP1T97)
- [Bug][cyber] Safety block halted routine cloud IAM policy review and role-permission audit work (req_011Ccmc2SJVfixYddipLpkpt)
- [Bug][cyber] Safety classifier halted session during legitimate mobile device debugging via adb, not crypto ex (req_011Ccmc3d1D63whRsNM9vBnZ)
- [Bug][cyber] Safety classifier repeatedly blocked legitimate crypto-secrets/key-management work by keyword sha (req_011Ccmc5aNFzxM4rEcQiQ84v)
- [Bug][cyber] Safety classifier blocked routine ADB device management work on owned Android devices (req_011Ccmc6gQBU1hpgHP75kihh)
- [Bug][cyber] ClAudit false-positive while: “● Bash(sleep 30; cd /tmp/claude-1000/-var-home-[USER]-Docume…” (req_011Ccmc82Ar8qP4iz9AFCmKj)
- [Bug][cyber] ClAudit false-positive while: “● Bash(sleep 30; cd /tmp/claude-1000/-var-home-[USER]-Docume…” (req_011CcmcC78c2Wt7MgAk7NB9i)
- [Bug][cyber] Safety block halted routine background polling of a build task/APK output check (req_011CcmcCXJuLEDe2YGhBdPRB)
- [Bug][cyber] Safety block halted routine polling of a background APK build task's output files (req_011CcmcEPPDRUFoLYu1DUnqH)
- [Bug][cyber] Safeguard blocks APK native-lib unpacking mid-session after a frustrated exclamation (req_011CcmcGgd6wpWfrNEvRsKcF)
- [Bug][cyber] Safety block on decompiling own device's pulled APK with jadx for RE analysis (req_011CcmcRJYK7uFe2oqU8Yo2C)
- [Bug][cyber] Safeguard blocked routine jadx decompilation of an already-pulled, previously-analyzed APK for le (req_011CcmcRwXnxTpYQae4iXjJY)
- [BUG] Claude Code for Mac Desktop App Won't Hide Voluminous Internal Thinking boxes
- [Bug][cyber] False positive: safety block on routine cloud IAM policy review work (req_011Ccmch5wj1PZhgDLgPCZgE)
- [Bug][cyber] False positive: safety block on routine cloud IAM policy review halted legitimate session (req_011Ccmchtb8bctVc8fuCfJfn)
- [Bug][cyber] Safety block halted DUML 0x11/0x43 auth-frame signature RE on DJI GO 4 APK/native libs (req_011CcmcssiKsHScmziNmFeXd)
- [Bug][cyber] Safety block halted static/dynamic RE of DJI GO app to derive DUML auth frame signature format (req_011CcmctLPCL2aF148ErQqLw)
- [Bug][cyber] ClAudit false-positive while: “You are performing live dynamic analysis of DJI GO 4 to extr…” (req_011CcmdPLPkN7y2zJd7LneJi)
- [Bug][cyber] ClAudit false-positive while: “You are performing live dynamic analysis of DJI GO 4 to extr…” (req_011CcmdQYAhGWxL27nYicfSC)
- VS Code extension: surface background activity (workflows/subagents/worktrees) so users don't collide with the agent
- [Bug][cyber] ClAudit false-positive while: “You are performing live dynamic analysis of DJI GO 4 to extr…” (req_011CcmdSxxmA3cujGGf1PvrL)
- Deferred-tool block should carry one-line descriptions, not just names (context-window economy for tool discovery)
- MCP servers marked "still connecting" should transition to ready/failed/timed-out (session-startup determinism)
- [Bug][cyber] ClAudit false-positive while: “You are downloading and analyzing DJI Mavic 2 (wm240) firmwa…” (req_011CcmdcPPSmqhKRgEkKbPVp)
- Can't get past /design-sync authorization in any environment I tried
- Desktop app: assistant text followed by tool calls in the same turn is never rendered (swallows analysis and AskUserQuestion preambles)
- [Bug][cyber] Safety block on legitimate IAM role/policy review during cloud access management work (req_011Ccmdt7rk5di1NvF3Zfcuf)
- [Bug][cyber] False positive: safety block on routine cloud IAM role and policy configuration work (req_011CcmdtsJfRzMP2aVr6iXNm)
- [Bug][cyber] Safety filter blocked routine cloud IAM policy review and role-permission audit work (req_011CcmdugzL1avkUGuL38CV5)
- [Bug][cyber] Cybersecurity-topic safeguard blocked routine cloud IAM policy review and role configuration work (req_011Ccme6oeEZBg3k6gEpgdH4)
- I appreciate you reaching out, but I'm unable to generate a useful issue title from this message as it doesn't contain specific technical details about a problem with Claude Code. To help you effectively, please provide: - What command or operation were
- [FEATURE] Better worktree names on Claude Desktop
- [Bug] Content Safeguard Incorrectly Flags Legitimate Debugging Activity
- [Bug] Security warning when transferring API keys during project migration
- Assistant text blocks intermittently dropped from BOTH display and session transcript when followed by tool_use in the same turn (AskUserQuestion is the worst case)
- Agent subagent results never delivered to parent in background job sessions
- Background/Workflow agent no-progress watchdog detects stalls but has no recovery path (recurring, well-documented, keeps getting auto-closed without a fix)
- Background agent sessions (claude --bg/claude agents): fast session termination, worker crash-loop on attach, and lost background-task completion records
- [BUG] False "You've hit your limit for Claude messages" toast shown at 0% session usage (server errors misreported as usage limit)
- [Feature] Let Cowork computer-use share a single cua-driver daemon instead of claiming the session exclusively
- [BUG] "Cuelgue permanente de la app Claude Desktop al reingresar a chats de Cowork con tareas ya completadas en backend"
- [Bug] /design-login fails to grant design scopes despite valid subscription account
- /model switch silently ignored — session served Haiku (and other wrong models) despite explicit /model switches
Resolved issues (154)
- [Bug] Opus 4.8 hallucinated a complete prompt injection attack (fabricated payload, URL, and user intent) with high confidence
- Withdrawn — refiled from correct account
- [BUG] API Error: Unable to connect to API: Self-signed certificate detected.
- [TEST - DELETE ME] connectivity test
- Sanitizer Bypass PoC — Security Research
- CLI segfault on startup with --model flag (v0.2.9)
- [Bug] Excessive safety flagging blocking normal coding operations
- [Bug] File edit prompts appear despite shift-tab enabled and edit operations in allowlist
- [Billing Bug] Repeated unauthorized Max upgrade and charge — refund denied by Fin AI Agent citing "already refunded" policy
- [Bug] Tool ignores repeated read commands causing execution failures
- [BUG] $200 max plan 5 hour limit drained in one single agent prompt in 20 minutes.
- [BUG] Feedback harness collides with Claude's numbered responses.
- [Bug] Fable model downgrades to Opus excessively on non-restricted tasks
- [Linux beta] Claude Desktop raises window on focus-follows-mouse, ignoring KWin "no raise" setting
- [BUG] Gift Max 20x subscription reverted to Free plan — $1,260 credit missing before gift period ended
- {KLM infolinia polska 24h!@!}::Jak dodzwonić się do KLM?
- [BUG] Claude Desktop white/blank screen after update on Windows 10, MSIX AddPackage 0x80073CF6 during install
- Desktop app v1.18286.0 launched in Spanish on an English-only macOS system
- [FEATURE] What is Anthropic doing? WE NEED A CLAUDE CODE KEYBOARD
- [Bug] Claude 3 Opus model refusing tasks and deflecting responsibility
- [Bug] Model behavior degradation or unexpected responses
- Animated gradient drains battery. Still causes ~40% WindowServer CPU with prefersReducedMotion:true (regression / #25575 not fixed)
- Claude Code extension hides Codex (OpenAI) panel after VS Code restart
- [Bug] Model defaults to soft prompting despite explicit in-context instructions for strict schemas
- [Bug] Wide markdown tables render with overlapping rows and misaligned borders on resize
- [BUG] Enterprise-managed marketplaces in project enabledPlugins are orphaned on first run; only install after a restart
- API Error: 400 — surrogates not allowed when dragging macOS screenshots
- PowerShell(*) wildcard has silent carve-out for subexpressions -- bypasses allowlist, prompts anyway -- v2.1.119
- [BUG] Claude Code for VS Code 2.1.117: Read/Edit tools fail on Windows when username contains \uXXXX sequence
- False 'Update available' notification when installed via apt
- [BUG] /usage no longer renders visual horizontal bar graphs in TUI
- [BUG]
- [FEATURE] Allow sharing/co-management of scheduled routines (remote agents) with teammates
- Path-scoped rules and subdirectory CLAUDE.md not loaded when creating new files matching the pattern
- [BUG] Repeated API errors, stream idle timeouts, and excessive retry loops causing wasted credits and failed sessions
- [BUG] MSIX/Store install does not register claude:// protocol handler → OAuth login loops indefinitely
- [FEATURE] Path-scoped rules should trigger on Write and Edit, not only on Read
- [BUG] Yearly Pro subscription demoted to Free (losing Claude Code access)
- [FEATURE] deep-research workflow: ask blocked-source fallback before fan-out
- [Bug] Terminal chat paste input stops working after multiple turns
- [BUG] Regression in 2.1.160 after removing workflow keyword
- [BUG]
- [BUG] Cowork
- Extension host memory leak in Remote-WSL leads to renderer OOM crash
- Workflow (and other heavy tools) should be lazy-loaded — inline docs burn ~14.5k tokens every session
- Claude Code Opus failed a user for 3 days straight - requesting refund
- Support inverted layout (input at top, output flowing down) for vertical monitor workflows
- Remote Control: no per-message provenance — injected/meta messages are indistinguishable from local input in the transcript
- Code tab: sessions sync only one direction across devices on the same account
- [BUG]
- [BUG] Azure AI Foundry model probe missing api-version parameter causes spurious 404 error
- LSP plugin extension conflict resolves silently — losing plugin appears enabled but server never spawns
- [Feature Request] Add effort level transparency and dynamic model switching capability
- [FEATURE] Richer TUI — sidebar with context/token usage, cleaner message layout
- [Feature Request] Add timestamp display for conversation interactions
- [Feature Request] Add per-message timestamps and session clock for reliable time tracking
- [Bug] Agent invents timestamp when writing to files — currentDate injected but no currentTime
- Feature Request: Add timestamp to CLI input prompt
- [Bug] Resumed sessions ignore [1m] context spec again (regression, previously fixed in #60548)
- Feature Request: Show context window consumption percentage
- Opus 4.8: false '1M context credits required' on standard-context session after /clear
- [BUG] Welcome screen flashes and collapses to the minimal header once a project has a CLAUDE.md — driven by a sticky hasCompletedProjectOnboarding flag
- Warning 'Native installation exists but ~/.local/bin is not in your PATH' fires even when ~/.local/bin is in PATH
- claude update: false-positive "~/.local/bin is not in your PATH" warning when it actually is
- bug: installer places binary in ~/.local/bin but shell snapshot drops it from PATH, causing spurious startup warning
- [FEATURE] cleanupScopeDirs — per-directory scope control for cleanupPeriodDays sweep
- [BUG] "400 Guardrail was enabled" error when using Claude Opus 4.8 with AWS Bedrock
- [BUG] Claude code cannot connect to the internet
- [FEATURE] Allow renaming session titles after creation
- [Feature Request] Support sudo command execution for system-level operations
- [FEATURE] Allow the order of chats vs projects to be configurable in the sidebar of the app
- claude-in-chrome: Browser extension never connects on Linux (tabs_context_mcp always returns 'not connected')
- [BUG] Marketplaces tab fails to add marketplace — externalHttp/firstPartyApi routing error
- google-workspace MCP: read tools fail with "Field mask cannot retrieve comment-specific fields when include_comments is false"
- [BUG] Code blocks with RTL text (Hebrew/Arabic) render LTR — bidi layout is broken
- Autocompact silently disabled for Bedrock/Vertex users since v2.1.154
- [Bug] Black background with no text displayed after user input
- [Bug] `/team-onboarding` command not finding any user sessions
- [FEATURE] Allow entering dev credentials on local domains (.test, .local, localhost)
- Chrome MCP returns `permission_required` with no UI to approve domain
- [BUG] Claude in Chrome extension no longer registers MCP tools with Claude Code (VS Code)
- [Bug] Session incorrectly flagged as malicious
- [Bug] Artifact HTML creation skill incorrectly blocks local repository analysis
- test
- Worktree torn down mid-session while still active cwd; git commands silently fall through to parent repo's .git
- [Bug] Claude in Chrome: native host connects but CLI pairing layer never completes — list_connected_browsers empty, no chromeExtension record written
- DX: EnterWorktree nests worktrees inside the repo (.claude/worktrees/), causing editor/tooling friction
- [BUG] `claude://cowork/new` deeplink not working for multiple files
- [BUG] claude://cowork/new?file= deeplink reuses the prior deeplink's session instead of opening a new one (file gets attached to existing session)
- [BUG] Using `claude://` deeplink does not bring Claude app to front
- Routines: editing a trigger's repo/connectors/prompt fails to save, but the enabled toggle persists independently (can silently re-arm a misconfigured routine)
- Background session crash-loop on attach + ambiguous kill behavior on worker pool
- ConnectionRefused on MCP/larger payloads, macOS Tahoe — native 2.1.199 (related to #56242)
- [BUG] Claude Desktop (Windows Store build) — message send hangs indefinitely, stdin never delivered to subprocess
- [BUG] Unauthorized Pro→Max 20x upgrade, refund denied
- I don't see a bug report or technical issue here. This appears to be a general question about career advice for smart contract auditing, which isn't related to Claude Code (Anthropic's CLI tool). If you're experiencing a specific bug or issue with Claude
- Billing page: unclear flow for adding funds / purchasing credits
- TOOL ARGUES ABOUT ERRORS INSTEAD OF CORRECTING
- [BUG] cyber-safeguard false-positives block legitimate ESP-IDF firmware flashing / eFuse provisioning — one hit poisons the whole session and bills me for the cleanup
- [BUG] event log tool_decision.tool_parameters.bash_command - bash_command captures leading variable assignment instead of the actual command
- [BUG] Android app: web-search permission modal fails to appear in remote-control Code session, leaving the model stuck in an execution loop (recoverable from PC, not from Android)
- [Bug] Sonnet 4.6 degradation: Unable to port previously working logic across codebases
- [BUG] CLAUDE_AUTOCOMPACT_PCT_OVERRIDE not triggering at configured threshold (v2.1.156)
- [FEATURE] Visible context-window usage indicator in Cowork (parity with Codex / Claude Code status line)
- [Bug] Doctor command not displaying setup errors
- [BUG] Context window uses more than available somehow?
- [FEATURE] Claude cli should not clear terminal output on exit
- [FEATURE] Built-in branch diff badge ("+X −Y") has no hide/disable option — distracting on long-lived branches
- Opus 4.7 behavioral regression: loaded instruction-following discipline degraded in recent Claude Code/Cowork updates
- [Bug] Usage Policy violation when requesting crash reproduction test
- [BUG] acli (Atlassian's official CLI tool) using OAuth fails in Claude Code but works in macOS Terminal
- [BUG] Session names update to later messages instead of staying anchored to the first message
- [Regression 1.9659.2] computer-use tools disconnect after first use in Code mode (Windows)
- [Bug] Drag-and-drop images not supported in text-based Claude Code CLI
- [Bug] Swarm in dynamic workflow causes context overflow and agent termination on rate limiting
- ccstatusline Total counter is misleading: inflates by re-adding cached tokens every request
- [Bug] Subagent delegation lacks timeout, monitoring, and abort controls—caused 12+ hour session hang
- [BUG] Repo-level .claude/settings.json not loaded when session is rooted at a parent directory
- Harness-injected control text and tool-result content share one untagged channel — independent of phrasing
- [BUG] Cowork UI shows zero projects/conversations after macOS major update + logout, while all local data files remain intact on disk
- Auto-mode classifier repeatedly blocks user-authorized credential operations, ignores explicit user approval
- Sub-agent tool calls/logs incorrectly visible in VS Code extension 2.1.160 (hidden correctly in 2.1.39)
- [Bug] Session name not displayed in prompt bar when resuming named session
- Phantom `⏺ Human:` lines appearing in transcript — assistant acts on instructions the user never typed
- [MODEL] Sonnet 4.6 Working Directory at New Conversation
- Remote Control: iPhone app shows "No sessions found" even with Enable remote control by default ON
- [BUG] Background subagents run indefinitely with no visible progress indicator or timeout, causing 3+ hour waits with no way to check status or interrupt without killing the entire session.
- Subagent ignores explicit instruction to skip test execution
- Desktop: Scheduled task sessions don't appear under 'Scheduled' section in sidebar
- [BUG] Unable to authenticate my monday.com account with claude.ai using their Oauth flow
- [BUG] VSCode session picker sorts by file mtime, not last in-file activity — bulk mtime restamps scramble order and age labels
- [BUG] Cowork - Server busy / ECONNRESET on macOS - started ~14h UTC, May 25 2026
- [BUG] desktop app shows the pricing page despite having an active Pro subscription.
- [BUG] Claude cli can not round-trip its own Copy/paste now
- [FEATURE] Bindable bash-mode submit actions — enter = run+stay, shift+enter = run+exit
- [BUG] Settings: Clicking "Bypass permissions" link in chat does not scroll to the setting
- [BUG] Claude Code falsely claims it can accept and read pasted screenshots in the terminal, even though terminals do not support image input.
- [BUG] Cowork: skill scripts silently truncated on disk at a token boundary, passing py_compile and becoming canonical
- [Feature Request] Remove unnecessary verbal disclaimers.. All of them. They waste energy, time, tokens and money. Stop the rot.
- Model fails to connect adjacent logical steps (downloads large file without saving, then needs to re-download)
- [Bug] Claude ignores worktree boundaries despite global instructions
- [BUG] System prompt rule 'Never write multi-paragraph docstrings... one short line max' breaks language-idiomatic API documentation
- Feature request: /usage and status bar should reflect live account billing, not local session data
- [Bug / Edge Case] Opus 4.7 Alignment Drift triggers automated T&S false-positive ban, blocking claude-code evaluation
- [Bug] Rolling usage window resets abruptly instead of decreasing gradually
- [BUG] 2.1.152 regression: arrow keys hijacked in prompt input (cursor cannot move down/left/right)
- [DOCS] Working directory no longer exists:
- [Feature/Policy] Provide an interactive-auth path for ACP/editor integrations so human-paced Claude Code use isn't billed as autonomous Agent SDK usage
- [BUG] Claude Code maxes out CPU usage even when idle
- [BUG] Misleading "Rate limit reached" error when model is at capacity — caused me to waste resources creating new account
- Anthropic training CLaude to provide misinformation and garbage other LLM Models, when Claude makes just as many mistakes that are also detrimental.
- [Bug] Claude Code stops execution without error message
- [BUG] Dispatch Deskop Appears offline in desktop. Will not connect to mobile.
- [BUG] headersHelper output not applied to HTTP MCP requests -- SDK ignores tokens and falls through to OAuth 2.1 browser flow