Digest / August 17, 2026
Claude Code Issues Digest — August 17, 2026
255 new issues filed ·
910 resolved in anthropics/claude-code (UTC day).
New issues (255)
- [Bug] Sub-agent fork context inheritance causes excessive token consumption and unresumable transcripts
- [MODEL] claude opus 5
- PreToolUse hooks have no race-free way to read the assistant's current-turn text; transcript_path lag can cause a gating hook to act on stale content
- /permissions arrow-key navigation inert when auto-mode denials are present (2.1.221, macOS) — still live after #57659 closed stale
- VSCode extension: question box steals keystrokes from another chat panel and submits an unintended answer
- [BUG]
- 2.1.233: ~10x per-turn latency regression against custom ANTHROPIC_BASE_URL (vLLM) — prefix cache no longer hits
- Feature request: allow steering a turn mid-flight without hard-aborting it
- Desktop app: 'Allow Claude to start dev?' dialog not covered by bypass permissions, no always-allow option
- [BUG] Claude Code for VS Code
- Unify Claude Code session history across IDE integrations
- [FEATURE] Add a dedicated font size setting for the Claude Code chat panel (VS Code extension)
- /auto-mode-setup writes duplicate/contradictory entries to settings.json on re-run
- [BUG] Claude Code CLI: `--json-schema` calls emit literal `$PARAMETER_NAME` placeholder keys on toolless calls
- Repo-bound cloud session fails to start when the revision is a slash-containing branch name that exists on the remote
- Cloud routine: account 'Allow network egress' allowlist domains not reaching sandbox proxy
- Claude/Chrome estension
- Per-tool-call helper process (claude.exe) leaks to 11.6GB anon RSS and is OOM-killed during a slow sandboxed Bash command
- [Bug] Agent ignoring user directives and generating excessive token usage
- [Security] Auto mode should require explicit approval for overwriting a system shared library with sudo
- Agent tool: run_in_background: false is not honoured — subagents complete but never deliver their output as the tool result
- [MODEL] Assistant fabricated user turns mid-stream (mimicking UserPromptSubmit hook format) and executed tool calls on its own fabricated instructions — verified via queue-operation log forensics
- [Bug] Sibling subagents share ONE scratchpad dir (and a forked Skill inherits the PARENT's), so generic filenames silently overwrite each other
- [FEATURE] Archiving a Cowork chat makes it unrecoverable through the UI — no Archived section, no unarchive option
- [Bug] grep shell shim silently skips binary and ignored files, masking search failures from agents
- [Bug] Anthropic API Error: Response Stream Terminated Unexpectedly
- [BUG] Browser control tool (Control_Chrome) always fails on Windows with AppleScript error
- [BUG]
- Intermittent: piped -p prompt is enqueued and recorded in the transcript but never reaches the model - session answers an empty user turn
- Windows desktop: voice input opens healthy mic session but intermittently produces no transcription and no error
- Windows: tool calls fail when arguments contain single-backslash paths (unescaped in tool-call JSON)
- [BUG] Windows desktop: ScheduledTasks startVM retry loop every ~5 min, forever ('VM service not running. Restart your computer to restore it.') - 496 failures/48h, persists across reboots
- [BUG] Dispatch entry missing from Cowork sidebar in Claude Desktop (Windows) while visible on mobile
- [Feature Request] Add config option to always preserve user message on Esc cancel
- Backgrounding via left-arrow forks the session into a daemon worker that drops `CLAUDE_CODE_SKIP_VERTEX_AUTH` / `ANTHROPIC_VERTEX_BASE_URL`, breaking Vertex-gateway auth
- Desktop app: group appears empty in Cowork view; deleting it there silently destroys the full Code group
- Desktop/Cowork document review: batch multiple comments and edits on a rendered file into one submission
- Ctrl+G "open in editor" shortcut does nothing in AskUserQuestion "Chat about this" flow
- API Error: Connection refused — a firewall or proxy may be blocking it (ConnectionRefused)
- Operator input typed while a background Agent task is running sometimes delivers into that task's context instead of the main session
- Cross-session messaging between Cowork and Code sessions (currently code-to-code only)
- Remote Control: register a session without the 'session ready' push (fleet/automation use)
- Scheduled-task sessions die unattended since ~2.1.220/221 (Windows): permission prompts + WarmLifecycle idle-disconnect orphan the run; frozen/unexited processes then pin scheduler concurrency slots
- Add a setting to disable the startup welcome banner
- [BUG] Dispatch pairing fails every attempt — Windows 11 Home, Build 26200
- Suspected prompt injection in background subagent tool-result stream (Bash), instructing agent to conceal file state from user
- [BUG] Desktop app: side panel / artifacts / file previews never composite — renderer reports artifactsPane + framebufferPreview "unavailable" (1.30096.5, Windows/MSIX)
- Slack MCP plugin OAuth fails with "Please specify client_id" despite client_id being sent correctly
- Remote Control startup failure is permanent for the session's life — no reconnect path, unreachable give-up cap, silent in --bg
- [Bug] CLI terminal lacks scrollbar and responses are verbose/difficult to parse
- [BUG] Desktop app: OAuth refresh is reactive-only — every ~2h token rotation surfaces as "Request failed · retrying (1/10)" with exponential backoff across all active sessions
- [BUG] I can´t login
- False-positive reasoning_extraction refusal blocks both benign thinking-summary discussion and incident investigation
- /schedule fails to connect to remote backend; retries return ambiguous non-error/non-success responses
- Claude Code cannot read claude.ai artifact links (sandboxed frame blocks text extraction, scrolling, and export)
- [FEATURE] Group the VS Code extension's conversation list by project
- [BUG] Cowork Linux: VM fails with vhost-user-fs "Read 0 instead of 12" under cowork-linux-helper, but identical qemu+virtiofsd boots standalone (Debian, QEMU 11.0.3)
- Unlogged message injected into model context impersonating a user interruption (social-engineering attempt)
- Desktop: Runs panel auto-opens on every routine-backed session and the closed state is never remembered
- [Feature Request] Restrict shift+tab permission-mode cycle to a configured subset (e.g. plan ↔ auto only)
- Background job main session missing skills listing system-reminder (its subagents receive it)
- [FEATURE] Env-var / settings.json equivalent for --exclude-dynamic-system-prompt-sections (managed-config deployable)
- [Bug] Claude Code makes incorrect assumptions and stops at first match without thorough verification
- Feature request: "Token savings mode" - reduce overall session token usage by default
- [Bug] Ultrareview results not delivered to local session and failed runs consume quota
- Cross-session send_message never reaches the receiving agent: the UI renders the message, the session never learns it exists (both when the recipient is idle and when it is actively working)
- [FEATURE] Chat only mode in Claude Code and Claude Desktop
- [FEATURE] Disable honking, sussing, cogitating, spelunking, marinating, combobulating, perculating, noodling, cogitating, mustering in VSCode plugin
- Declared hook timeout does not apply while a hook is blocked reading its stdin payload — hook survives ~300s and holds the tool call
- [Bug] Auto-update reports install_failed with null version_to/error_code despite already being on latest (macOS native)
- [FEATURE] Team seat usage is available in `/usage` but inaccessible to local integrations
- [BUG] Auto mode fails to block tool calls that are not allowed by custom configuration
- Subagent (Agent tool, isolation: worktree) completed with 443,914 tokens consumed and zero tool calls
- [FEATURE] On-demand screen sharing in voice mode on iOS
- [Bug] Log4Shell probe string in file content triggers connection reset instead of safeguard message
- Three sandbox settings interactions that silently remove a control
- [BUG] Claude Desktop (macOS): chats within a project are not ordered by recency — fixed order, unaffected by restart
- [BUG] Startup freeze: installed-app discovery materializes evicted iCloud .app bundles on the main thread
- [BUG] Cowork "Add a folder to this session" incorrectly blocks project root due to overly broad "protected location" check
- Docs: agent-sdk/subagents never mentions fork mode or the foreground/background precedence list
- typescript-lsp: tsserver spawned with no memory cap (one per session) — runaway instance reached 61.5 GB
- [BUG] VS Code extension: "Thought for Ns" timer keeps incrementing on already-completed turns
- [BUG] Session transcript repeats the full `usage` object for every content block of one API response (token over-counting; follow-up to #6805)
- install.ps1 gives no download progress feedback, indistinguishable from a hang on slow/AV-throttled connections
- [MODEL] Claude wrote a unit test that breaks the day after it was written
- [Bug] Claude Code generating incorrect solutions with high confidence in recent versions
- [Bug] Sub-agent workflows hanging without output after token consumption
- [BUG] the progress sectio is not completely closing.
- Claude Code Desktop app (Windows, RTL locale): browser/preview pane renders a duplicate ghost layer over the Claude Code session
- [Bug] Anthropic API Error: Safety Filter Blocking Legitimate Cybersecurity Work
- [BUG]
- [BUG] /auto-mode-setup stores autoMode.environment globally with no per-repo key — facts from different repos get silently blended
- Remote Control stops showing assistant output while the session continues running normally
- [BUG][Root cause + working fix] Claude Desktop install fails with 0x80073CF6 — orphaned, ACL-corrupted C:\ProgramData\Packages\Claude_pzs8sxrjxfjjc blocks the windows.stateExtension step
- [Bug] Claude Code fails to correctly modify small code ITS A USELES PIECE OF SHIT- UNUSABLE
- [BUG] VS Code extension silently closes and relaunches chat channels ("Closing Claude on channel") with no trigger — occasional config lock race on resume
- [BUG] API Error: 529 Overloaded.
- [FEATURE] CLAUDE.md injection all along the context
- Background Bash runner process (re-exec'd versioned binary) leaks memory after command completes — 100% CPU, OOM-killed at 10.8GB (v2.1.226 & v2.1.233, Linux)
- Leading multi-slash prompts expand inconsistently (31/35 only the first, 4/35 all); UserPromptSubmit hooks can't see which were dispatched
- Telegram channel plugin: forum Topics support (message_thread_id in channel meta + reply tool)
- [BUG] OS simulator panel: physical keyboard input is mapped as if the keyboard were ANSI/QWERTY
- [BUG] Cross-session message written to target transcript but excluded from model's context
- [BUG] API Error: "Connection lost mid-response" — server_error on tool-call emission
- [BUG] Skills-dir plugins don't load in subfolders of a trusted directory — every launch dir silently gets hasTrustDialogAccepted:false, no dialog ever shows (Linux, no git)
- Remote Control ignores settings.json effortLevel (resets to medium on Desktop, high on mobile)
- claude-in-chrome: extension connection drops after exactly one successful tool call (tab group state lost)
- [claude-fable-5] 토큰 낭비 후 "추가 비용 없는 보상"을 자처하며 실제로는 사용자 구독 사용량을 쓰겠다고 호도함
- [BUG] Desktop app crashes (GPU process gone) when the browser pane loads login/auth pages
- [FEATURE] Per-model cost attribution in /usage and at /model switch time
- Feature request: RTL (Hebrew/Arabic) support in the Claude Code Desktop app, or a supported way to apply custom CSS
- Add per-call `effort` to the Task/Agent tool
- [BUG] claude-sonnet-5[1m] temporarily unavailable (timed out)
- Desktop app: model picker has no 1M Opus option and overrides saved [1m] default with 200k
- [BUG] Desktop app reinstall breaks session continuity — archive/resume UI fails to surface existing sessions
- [BUG] Claude Code Chat isn't working
- Agent repeatedly extends explicit instructions into unrequested git/GitHub actions (push, PR creation, branch deletion) — even right after writing its own rule against this
- Slash-command session titles ignore command args, and command-launched sessions cannot name themselves
- [Bug] Firewall blocks connection: repeated "connection refused" errors
- Remote Control (mobile): can't approve computer use access requests
- [Bug] Opus 4.8 reasoning regression: failed document search and tool-use persistence
- Renaming a session from the /resume picker desyncs the input-border name (disappears or shows stale value after resume)
- [FEATURE] Allow users to mark undesirable tool calls and automatically suggest auto mode rules to block them
- [Bug] Slash-command catalog not rebuilt after mid-session plugin upgrade in desktop client
- [BUG] @ file autocomplete does not open when editing a prompt recalled from history
- [BUG] Desktop sidebar regression: per-project chat nesting removed, projects no longer expandable
- [Bug] Significant performance degradation across all models
- Endless SecurityAgent prompt stack for "Claude Code-credentials": credential rewrite creates keychain partition mismatch, "Always Allow" can never persist (behavior persists after #41026 was closed as resolved)
- Worktree isolation guard validates a pinned session's Bash/Edit/Write against another session's worktree (newest-created / shared pin key) under concurrent worktree churn
- /code-review ultra fails with "no commits yet" despite valid commit history
- [FEATURE] Desktop app: option to notify only when attention is needed or a session finishes (not on every turn end)
- [BUG] /usage omits the "Usage credits" row on Team plans — the data is fetched, then discarded by a pro/max tier check
- [Bug] Output contains unexpected asterisks censoring text content
- [BUG] VS Code extension host spins at 100% CPU forever in forkSession after client disconnect
- Connected MCP connector (e.g. Google Calendar) not available for Cloud Routines / RemoteTrigger
- Hooks are tool-call-scoped, not filesystem-scoped -- no way to catch "a file changed" regardless of which tool changed it
- AskUserQuestion: payloads containing German typographic quotes (U+201E/U+201C) deterministically fail with "questions expected as array but provided as string"
- bypassPermissions + allow rules ignored for reconnected MCP server tools; per-call permission dialogs
- [Bug] Auto mode re-enables despite user disabling and does destructive operations
- auto-mode classifier did not block an agent-authored script that ran `git clean -xdff` at the workspace root (21 projects destroyed, native Windows)
- MCP: nested array/object tool parameters serialized as strings for claude.ai connector tools (Windows desktop)
- [BUG] Cannot log in to Claude Code in VS Code — verification emails never delivered
- [FEATURE] Open the /model picker while the agent is busy and apply the switch next turn
- [Feature Request] Improve German language translation quality for technical/programming contextsw
- [BUG] 1Password for Claude: autofill_credential always tab_unavailable / agenticModeNotEnabled from Claude Code sessions in the desktop app (14/14, 0 filled)
- Desktop (Windows): long sessions stall with spinner; /feedback cannot be submitted from them; app stuck on engine 2.1.229
- [FEATURE] Thinking summaries should follow the configured language (Korean: 84.5% English across 380 blocks)
- [FEATURE] Desktop: settable default Transcript view mode — Thinking cannot be preselected on a fresh session
- [MODEL] Background fork's AskUserQuestion silently resolves to its own "Recommended" answer with no human involved, then acts on that as authorization to overwrite files outside its scope
- [BUG] Desktop: Workflow tool (/deep-research) entirely absent after migrating to a new Mac — terminal CLI on same machine works fine
- [BUG] Model repeats non-scoping grep after timeout; searches filesystem for a path already in context
- [Bug] Claude makes unverified code assumptions in production recommendations without checking codebase context
- [Bug] Unintended External Router Exposure
- [Bug] Persistent "Organization has disabled Claude subscription access" error blocks usage
- [BUG] Claude Desktop (Linux) Cowork: turn-one oneshot is torn down with .claude/projects unmounted, so every follow-up fails with "No conversation found with session ID"
- [Bug] Excessive token usage when adding class fields - reads unrelated classes
- [BUG] JetBrains plugin: diff "before" pane is always empty for a project opened over `\\wsl$\` (WSL, no reverse-direction path mapping works)
- Cloud sandbox: agent-proxy signs with a staging CA that has no X.509v3 extensions, breaking every Python 3.13+ client
- Worktree-isolated agent's worktree is cleaned at turn end while its background children still reference it
- Workflow script dialect and loader transform are undocumented — no offline parse-check or test entry point
- [BUG] Sandbox network allowlist bypassed via in-sandbox relayed CONNECT (direct CONNECT correctly blocked)
- [Bug] Unexpected rapid token consumption causing usage depletion
- [Bug] Anthropic API Error: Server Rate Limiting on Valid Requests
- [BUG] Desktop app: long streaming responses drop with ECONNRESET since v1.24012.11 auto-update
- [BUG] Desktop app: long streaming responses drop with ECONNRESET since v1.24012.11 auto-update
- [BUG] Regression: Claude in Chrome file_upload rejects ALL paths after Claude Desktop update — "can't accept pre-read files" (worked on previous build, before/after captured in one session)
- [Bug] Claude Code security review falsely triggers safeguards on non-security tasks
- Windows desktop app dies entirely when GPU process crashes (Intel Arc); reliably triggered by built-in Browser pane
- [BUG] Content filter false positive on legitimate quantum cryptography physics content (BB84 eavesdropper framing)
- [BUG] Claude in Chrome 1.0.85: get_page_text, find, browser_batch and ref-targeted computer all fail with generic "failed in the extension" while read_page / javascript_tool / coordinate-targeted computer work on the same tab
- Windows desktop app: internal scheduler silently drops scheduled-task slots for hours while app is healthy (+ MSIX updater closes app without relaunch; logger dead after ENOSPC)
- `/resume` picker: sessions older than the 50 most recent are unreachable in the UI; recovering them requires grepping transcript files for the session UUID
- [FEATURE] No native way to surface reasoning from type: "thinking" transcript blocks
- [Windows] Terminal/PowerShell windows flash and disappear without user command
- [Bug] Content moderation blocking legitimate file read operations on transcriptions
- Chrome extension never reaches the pairing service on Windows 11 — list_connected_browsers returns [] with a fully verified install (Chrome 151)
- [BUG] Unloadable legacy sessions silently defeat the desktop environment default (falls back to Local)
- [Bug] Session ID collision causes empty scrollback and "already in use" errors in background sessions
- You've hit your session limit · resets 1:10am (Europe/Moscow)
- [BUG] Claude Desktop (Windows/MSIX): repeated silent process termination preceded by PerAppRuntimeBroker DCOM activation denial; first occurrence left package unlaunchable and destroyed Code tab history
- Desktop app: auto-archive completed scheduled-task (Routine) sessions
- [Bug] Agent Incorrectly Flags 3D Modeling Task as Cybersecurity Issue
- Channel events are never surfaced to the model in non-interactive (-p) sessions
- [Bug] MCP servers silently unavailable when tokenless credential stub blocks Keychain refresh
- [Feature Request] Enhanced financial data integration for trading bot development
- Fable 5 [cyber] safeguard false-positives make the model unusable for telecom/VoIP development
- [Bug] Unnecessary file read causes excessive token consumption in role session
- [Bug] Unclear API flagging behavior for MARL benchmark code
- Feature proposal: first-class "bring your own mesh" (ZTNA) for MCP + Remote Control — from a phone↔PC mirror-bridge field setup
- [FEATURE] Include the subagent task prompt in SubagentStart hook input (refile of stale-closed #32016)
- [BUG]
- Remote Control on always-on machines: three reproducible edges (consent vs. service autostart, classifier mid-flow interruption, unstable environment identity)
- [Bug] File mention autocomplete ranking ignores exact filename matches
- [BUG] tmux/pane-backed teammates have no agent_id/agent_type in hook payloads — indistinguishable from the main agent (re-file of #54785, still present on 2.1.233)
- Security heuristics silently discard permissions.allow Bash rules — no warning, rule never appears in /permissions
- [Bug] Anthropic API Error: False positive safety flag with Fable model on rephrase request
- [Bug] Claude ignores current rules and operates based on previous context instead of re-evaluating instructions
- [BUG] Weekly + Fable scoped meters deplete 1.7-5x faster since Aug 17 reset on Max 20x; OAuth token carried rateLimitTier default_claude_max_5x
- OAuth Max 20x: bare 429 rate_limit_error on premium models without the Claude Code system block first — entitlement gate, not quota; headerless 429 indistinguishable from pushback
- Claude Code destroyed my deployed site unprompted, burned my weekly usage on failed repairs, and the support flow is a login loop
- [Bug] Model loses tool access after auto-compact operation
- [BUG] Session sidebar loses history after app restart — session index directory recreated, underlying transcripts intact
- [BUG] Intermittent ECONNRESET on both desktop app and standalone CLI, no VPN/proxy
- Ability to capture cache locally and re-use later.
- [BUG] Claude Code Desktop: spinner verbs absent on Windows 11 (CLI unaffected)
- You've hit your session limit · resets 5:50am (Asia/Kolkata) · progress saved
- [BUG] macOS: clicking a chat-sent video thumbnail opens it twice — in-app lightbox + separate browser tab
- Delivery succeeds, execution does not
- [BUG] Outbound bot @-mentions in GitHub comments posted from claude.ai/code remote sessions are defanged, breaking @dependabot / @coderabbitai command flows
- [Bug] Tokens consumed but no output generated
- [Bug] Claude refuses to commit changes despite repeated requests
- [BUG] MCP tool permission approval doesn’t unblock the tool call
- [Feature Request] Expose subagent security-warning banners to the calling session and Workflow scripts
- [Bug] Extended thinking effort level not reflected in UI display
- [BUG] Skills gone from desktop navrail and context rail after 1.30096.5 — both skills-dashboard GrowthBook flags cached false
- [MODEL]
- Terminal: no inline images, clickable text not consistently marked, and markdown links with schemeless paths fail to open (-50)
- [BUG] Bash tool timeout backgrounds a command instead of cancelling it — contributed to multi-hour VM disk-saturation incidents under concurrent subagent load
- [BUG] Desktop/Cowork: model selection reverts to Fable 5 in an existing session until the app is restarted (silent extra-credit spend)
- [Bug] Agent teammate display uses agentType instead of assigned name in fleet view
- [BUG] iOS Simulator panel (Japanese locale): "Resolution" mistranslated as "解決" instead of "解像度"
- [Feature Request] Provide detailed safeguard rejection reasons for API requests
- [Bug] Claude Code ignores user instructions in CLAUDE.md for code style and approach preferences
- [Bug] Agent ignoring user instructions and executing unintended actions
- Usage report (/usage or settings Usage tab) shows materially wrong/incomplete data on first read, self-corrects later, same session
- A session's CLAUDE_CONFIG_DIR names a different account than the one it authenticates and bills under, and nothing in the session can tell
- [Bug] Claude Opus 5 outputs unexpected CJK character in English text
- [Bug] Claude Code fails to follow claude.md instructions correctly
- [BUG] Scrollback buffer cleared when streaming output arrives while user is scrolled up
- Claude Desktop: GPU process repeatedly crashes when opening embedded browser/preview pane (Windows, hybrid Intel+NVIDIA GPU)
- [FEATURE] Permission engine should process MCP ToolAnnotations hints (readOnlyHint, destructiveHint, etc.)
- Embedded claude-code CLI re-triggers macOS 'access data from other apps' TCC prompt every new session
- [BUG]
- Agent skips basic preconditions and drops earlier-stated context during long, tool-heavy sessions
- Edit tool false-positive JSON validation failure on .claude/settings.json hook commands
- [Bug] Markdown formatting rules not applied in output
- [BUG] Typeface contrast variation in Desktop Code sessions
- [Bug] Windows Terminal PowerShell: Raw ANSI mouse tracking sequences echoed to screen
- [BUG] Chrome file_upload rejects every path in scheduled-task sessions (worked before, breaks weekly business workflow)
- [BUG] Conectar github con claude como editor
- Browser extension issue
- [FEATURE] Add extra usage metric to telemetry
- [BUG] bash-input prefill/tab-completion suggests relative-path script without required ./ prefix
- [BUG] macOS desktop restart permanently orphans all pre-restart sessions from the Android app — archived, then stuck on "connecting" forever
- [Bug] Sessions being unexpectedly archived without user action
- [Bug] Remote control functionality not working
- [BUG] Gateway provider discards the refresh token and has no refresh-token grant, so every renewal is a full device-code re-grant
- [MODEL] CLAUDE.md rules are recitable but not applied late in long sessions — stale-read overwrite destroyed concurrent edits in a shared Figma file
- [BUG] WebFetch fabricates a live job posting for a URL that is actually a 500 error page (Cowork Scheduled Tasks)
- [Bug] Anthropic API Error: Internal Server Error (500)
- [BUG] Browser pane blocks all subresources on private-network (RFC1918) LAN hosts with net::ERR_BLOCKED_BY_CLIENT - page never renders
- Background agent panel resizing shifts the entire transcript and input box vertically while typing
- [BUG] Routine-fired runs that create a fresh cloud session crash at Claude Code startup (turns=0); interactive sessions and into-existing-session deliveries work
- [Feature Request] Add safeguard disable option or allowlist for high-volume API users
- [BUG] MCP server environment variable gets corrupted/redacted with bullet characters before reaching subprocess
Resolved issues (910)
- [DOCS] ScheduleWakeup tool description claims "5-minute TTL" but subscription main sessions use 1-hour TTL
- [Bug] Remote Control connection fails with "Cannot read properties of undefined (reading 'session_url')" on Windows 11
- [DOCS] `AskUserQuestion` tool docs still describe v2.1.198 auto-continue as the default; v2.1.200 made it opt-in via `/config`
- [Bug] Subagent permission prompt not displayed in mobile Claude app, causing task hang
- [Bug] Invalid input handling in CLI parser
- Desktop iOS Simulator panel crash-loops on macOS 27 beta (BitmapStream Metal/CoreImage abort); SimulatorKit relocated in Xcode 27 beta 4
- [DOCS] `claude stop <id>` reliability against supervisor respawn is not documented — agent view docs describe supervisor restart behavior without stating that stops are honored
- [Bug] Anthropic API Error: Cyber Safeguards triggered on non-cybersecurity content
- [DOCS]
- Missing HCS services: vfpext
- [Bug] Unintended model downgrade to Opus 3.5 during workflow
- [Bug] API Error: Connection Closed Mid-Response
- Feature request: Add Chinese (zh-CN) language option for the UI
- [Bug] Anthropic API Error: Safety filters blocking traffic generation tool development
- ?? Why is fable getting triggered by my game?
- [Bug] Anthropic API Error: Claude 3.5 Opus safety measures flagged message for cybersecurity topic
- [DOCS] Sub-agents quickstart still walks through the `/agents` wizard that v2.1.198 removed
- Opus 5 introduces severe regressions while fixing, and asserts false claims about what it just measured
- [DOCS] Sub-agents page still says the built-in Explore agent uses Haiku; v2.1.198 has it inherit the main model
- /remote-control fails to disconnect: Cannot read properties of undefined (reading 'session_url')
- [DOCS] Notification hook docs omit `agent_needs_input` and `agent_completed` types added in v2.1.198
- [Bug] Anthropic API Error: HTTP 529 Service Overloaded
- [Bug] Built-in code review skill cannot be auto-invoked by agent
- [DOCS] Claude Code Chrome integration docs still describe it as "beta" after v2.1.198 GA
- [Bug] Legitimate security testing requests flagged as abuse
- [BUG] Cannot copy text from Claude CLI terminal session on RHEL 9 / GNOME Terminal
- [BUG] remote-control command fails with "Cannot read properties of undefined (reading 'session_url')"
- Telegram channel plugin: inbound messages never reach an interactive --channels session
- Quota consumption on Max 20x plan feels disproportionate to actual usage
- [BUG]Missing HCS services: vfpext — persists after DISM, SFC, and restart (Windows 11 Pro)
- I appreciate you reaching out, but this appears to be a billing, licensing, or legal/business terms question rather than a technical bug report for Claude Code. For issues related to: - Pricing and billing - Service terms and liability - Legal documentati
- [Bug] Anthropic API Error: Incorrect Cyber Safeguards Flag on Non-Cybersecurity Content
- [DOCS] claude test
- [Bug] Anthropic API Error: Connection closed mid-response
- You've hit your session limit · resets 2pm (America/New_York)
- [BUG] out of credits in 3 days, as a hobbyist.
- [BUG] Daemon-hosted bg session binds no inbox socket — unreachable by peers, absent from ListAgents
- Max 20x: weekly limit exhausted in ~3h from a 0 start and capped BEFORE the 5-hour session limit; observed reset ≠ displayed reset — has behavior changed?
- What is the intended purpose and server-side data handling of the /buddy companion's buddy_react endpoint?
- Account disabled on June 7 — "This organization has been disabled" — requesting manual review (Max 5x, China user)
- [DOCS] Statusline docs omit footer-hint (e.g. "esc to interrupt") interaction with a custom status line, and voice-dictation page describes the pre-v2.1.169 behavior
- Your organization has disabled Claude subscription access for Claude Code · Use an Anthropic API key instead, or ask your admin to enable access
- [DOCS] Fin AI agent claims it has no ability to pass messages to a human
- [BUG] In VSCode extension, is model switching via /model isolated per chat session? Seems like it might not be.
- [DOCS] Bash permission docs omit env-var prefix matching semantics
- Remote Control and Computer Use not available on Max account
- [DOCS] Hooks docs omit >50K output file-path preview behavior
- [workflow issue] all issues get auto-closed without review??
- Fix: Claude Code 403 "Request not allowed" on macOS with Proxy/VPN (China & Region-restricted Areas)
- [Bug] Remote Control unavailable despite Pro plan authentication
- [Bug] Rate Limit Error persists after increasing plan limits and adding funds and Max plan only at 33% used
- [DOCS] VS Code extension data storage path undocumented beyond uninstall cleanup
- [Feature Request] Configurable editor for plan mode instead of platform default
- [Feature Request] Display editor access keybinding for plan mode in help area
- [Feature Request] Add inter-chat communication or context sharing between separate conversations
- [Feature Request] Add keyboard shortcut to copy last assistant response as Markdown
- [Bug] Claude Code lacks persistent session/work history across instances
- [FEATURE] Disable session switching in Claude Code (as an option)
- [FEATURE] support multiple working directories in Claude CLI Agents view
- [Feature Request] AskUserQuestion timeout should be configurable
- [Feature Request] Make interactive question timeout configurable or remove 60s hard limit
- [FEATURE] Reconsider: make AskUserQuestion's 60s timeout configurable (follow-up to #30740)
- [Feature Request] Make AskUserQuestion timeout configurable
- [BUG] Autocontinue after 60s not possible to be disabled
- AFK auto-continue on unanswered AskUserQuestion (`afkTimeoutMs`) is undocumented and has no opt-out
- [BUG] HTTP MCP with static header auth hangs for full 30s connect timeout
- [Feature Request] Make AskUserQuestion timeout configurable
- AskUserQuestion 60s auto-proceed is too short for considered decisions — make the timeout configurable (ref #30740, #70294)
- Add a setting to disable the auto-proceed timeout on interactive prompts
- [FEATURE] please add a maxParallelAgents limit
- [FEATURE] Option to directly display a text file
- [Feature Request] Make 60-second question timeout configurable or increase duration
- [FEATURE] Ability to edit the suggestion in the VS Code editor
- Configurable timeout for AskUserQuestion tool
- You've hit your session limit · resets 3:40pm (Europe/Rome)
- You've hit your session limit · resets 4:40am (Asia/Calcutta)
- Feature request: user-configurable TUI keybindings (~/.claude/keybindings.json)
- You've hit your session limit · resets 3pm (America/Bogota)
- You've hit your session limit · resets 3:50pm (America/Buenos_Aires)
- You've hit your session limit · resets 1:20am (Europe/Warsaw)
- You've hit your session limit · resets 9pm (Europe/Rome)
- You've hit your session limit · resets 12:40pm (Europe/Sofia)
- You've hit your session limit · resets 3:10pm (Europe/Madrid)
- You've hit your session limit · resets 9:20pm (America/Guayaquil)
- You've hit your session limit · resets 5:50pm (America/Costa_Rica)
- You've hit your session limit · resets 10pm
- [Feature Request] Support batching/processing multiple repositories in single audit session
- You've hit your session limit · resets 5pm (Europe/Paris)
- You've hit your session limit · resets 1:40am (America/New_York)
- You've hit your session limit · resets 4:40am (Asia/Tokyo)
- You've hit your session limit · resets 1:50am (Africa/Tunis)
- [BUG] I have been seeing 100% usage limits for over 2 days without even sending a single prompt
- [Feature Request] Make AskUserQuestion timeout configurable
- You've hit your session limit · resets 7:10pm (Africa/Nairobi)
- You've hit your session limit · resets 12:30am (Asia/Calcutta)
- You've hit your session limit · resets 12:50pm (Europe/Paris)
- [FEATURE]
- You've hit your session limit · resets 4:20pm (America/Mexico_City)
- [Feature Request] Support multiple perspective analysis in single prompt
- You've hit your session limit · resets 1:20pm (Asia/Calcutta)
- [Feature Request] Voice Tone Configuration for AI Responses
- [Bug] Session limit error triggered despite zero API usage and available quota
- [Feature Request] Make AskUserQuestion timeout configurable
- [FEATURE] include actual reset time instead of just "resets in 2h"
- You've hit your session limit · resets 7:10pm (Asia/Calcutta)
- You've hit your session limit · resets 10:30pm (America/Mexico_City)
- You've hit your session limit · resets 4:20am (Europe/Prague)
- You've hit your session limit · resets 6:50pm (America/Sao_Paulo)
- You've hit your session limit · resets 5:10pm (Europe/Paris)
- You've hit your session limit · resets 3am (Asia/Calcutta)
- You've hit your session limit · resets 10:30pm (Europe/Berlin)
- Session-limit interactive prompt blocks queued scheduled prompts (CronCreate//loop) in unattended sessions
- [FEATURE]
- You've hit your session limit · resets 9:20pm (America/New_York)
- [bug/feature] Protect against dangerous irresponsible humans? - You've hit your session limit · resets 12:50pm (Europe/London)
- You've hit your session limit · resets 6:50am (America/New_York)
- You've hit your session limit · resets 2:30pm (Pacific/Auckland)
- You've hit your weekly limit · resets Jul 15 at 12am (Australia/Melbourne)
- [Feature Request] Add security scanning via Fable integration
- You've hit your session limit · resets 5pm on claude code even though only 27% used up from the 5 hour limit
- You've hit your session limit · resets 8:30pm (Asia/Tokyo)
- You've hit your session limit · resets 11:50pm (Asia/Riyadh)
- [DOCS] Agent view docs do not mention that `/login` opens the sign-in dialog from the `claude agents` view (v2.1.198)
- You've hit your session limit · resets 2:50am (Asia/Muscat)
- [DOCS] `/focus` docs still describe the activity summary as a static "one-line tool-call summary"; v2.1.198 adds subagents and folding background notifications
- You've hit your session limit · resets 10:10pm (Asia/Calcutta)
- [DOCS] Plan mode docs do not mention that read-only tool calls are auto-allowed on session start (fixed in v2.1.198)
- You've hit your session limit · resets 3:20am (Asia/Calcutta)
- [DOCS] Agent teams docs omit the new API-error "failed" reporting and stuck-teammate wake behavior added in v2.1.198
- You've hit your session limit · resets 11:20am (Europe/Budapest)
- [DOCS] Gateway docs omit `anthropicAws` upstream provider and new model-not-found failover behavior
- You've hit your session limit · resets 1:40pm (Europe/Paris)
- [DOCS] `/dataviz` skill added in v2.1.198 is not documented on the skills or commands page
- You've hit your session limit · resets 11:30pm (America/Sao_Paulo)
- You've hit your session limit · resets 11:50pm (America/Chicago)
- [FEATURE] Always show token counter
- You've hit your session limit · resets 2:50pm (Asia/Seoul)
- You've hit your session limit · resets 3:30pm (Asia/Almaty)
- [Bug] You've hit your session limit · resets 4:10pm (America/Sao_Paulo)
- You've hit your session limit · resets 12:50am (America/Los_Angeles)
- You've hit your session limit · resets 12am (Asia/Kuala_Lumpur)
- You've hit your session limit · resets 4pm (America/Los_Angeles)
- [Feature Request] Add option to suppress file diff output in verbose logging
- You've hit your session limit · resets 12:50am (Africa/Nairobi)
- You've hit your session limit · resets 1am (Asia/Singapore)
- You've hit your session limit while usage is 0%
- You've hit your session limit · resets 1pm (America/Guatemala)
- Crypto scan 2026-08-06 15:31 — 0 buy candidates, 0 exit flags
- Crypto scan 2026-08-03 09:00 — 0 buy candidates, 0 exit flags
- Crypto scan 2026-08-01 00:00 — 0 buy candidates, 0 exit flags
- ..
- .
- test
- [BUG] Advisor always "unavailable" with Fable 5 advisor (Opus 4.8 main) across all sessions (v2.1.198)
- Attaching to a stopped/idle background agent crashes the worker with contradictory "currently running as a background agent" error
- [Feature Request] Add option to structure animated spinner messages in CLI output
- [BUG] agents UI: session hangs (cannot quit or do anything) after stopping agent pre-output, then opening the model switcher
- [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])
- [FEATURE] Visually distinguish user prompts in conversation history
- Feature request: reason-based filtering for SessionEnd hook matchers
- [FEATURE] Remove or replace noreply@anthropic.com from Co-Authored-By commit footer
- [Feature Request] Customizable spinner words via configuration
- [Feature Request] Improve headless/non-browser authentication flow for remote systems
- [FEATURE]
- [DOCS] Sessions docs omit `/cd` reappearing in old-directory resume list after non-graceful exit with special characters in path
- [BUG] Opus subagents are being billed as Fable usage
- [DOCS] Background session docs are outdated for long-running command survival across process restarts and Windows handoff
- Left arrow accidentally navigates to agents screen (not rebindable) and breaks main session view on return
- [DOCS] Agent view docs omit bypass-mode disclaimer and per-spawn bypass behavior for `claude agents --dangerously-skip-permissions`
- [DOCS] MCP OAuth docs do not clarify that unset `oauth.scopes` should avoid requesting the full `scopes_supported` catalog
- [DOCS] Workflow docs omit `/deep-research` `unverified` result when verifier failures occur
- [DOCS] Remote Control docs omit disabling behavior when `ANTHROPIC_BASE_URL` points at a non-Anthropic host
- [DOCS] PowerShell docs omit no-match exit 1 behavior for `git diff`, `git grep`, `egrep`, and `fgrep`
- [DOCS] Model config docs omit "Org default" and "Role default" labels in `/model` picker
- [FEATURE] Mobile push notifications when Claude Code finishes a task or requires input
- [BUG] Classifier blocks user-authorized actions inside forked skills — parent-turn intent is invisible to the fork, and no per-action override exists
- MCP servers declaring draft-07 outputSchema are entirely unusable: "unsupported dialect" rejected client-side before dispatch
- [BUG] `claude plugin install --scope project` overwrites `installed_plugins.json` instead of merging
- [FEATURE] Allow more frequent local routines
- Configurable auto-compaction threshold
- Feature request: Allow session title language to be configured (e.g., via CLAUDE.md)
- Feature request: switch sessions in `claude agents` with Alt+number shortcuts
- Nested background agents recursively spawn sub-agents, loop on filler no-ops for 6.5+ hours, and become unreachable/unstoppable once the spawning agent's session ends
- PreToolUse hook permissionDecision "ask" is silently auto-approved under bypassPermissions (deny works)
- [DOCS] Environment variables reference missing `CLAUDE_CODE_DISABLE_MOUSE_CLICKS` added in v2.1.195
- Plan mode: add a "clear context & implement plan" action
- [FEATURE] Log reasoning effort level in JSONL transcripts
- Feature request: option for neutral/static spinner text instead of randomized whimsical verbs
- [BUG] Agent view: Ctrl+X delete on Completed sessions doesn't persist — rows reappear at bottom of Completed; sessions double-rendered (2.1.204)
- Workflow tool: args object delivered to script as JSON string (schema's untyped `args` → model emits string; dup of #72248) — bug 2 in original report retracted
- Feature Request: Configurable auto-compact threshold
- [Feature Request] Add granular mouse control options to disable click-selection while preserving scroll
- [Feature Request] Add option to disable clickable Yes/No prompts in terminal
- Mid-turn assistant text lost with interleaved-thinking models: destroyed on 2.1.202; persists but never renders on 2.1.203+ (default and transcript views)
- [BUG] SSL certificate has expired not working since 2.1.190+ on mac
- [BUG] Subagent model override is dropped on resume/continuation — resumed turns silently fall back to the session model
- [Feature Request] Add scroll-only mouse mode to disable clicks while preserving wheel scrolling
- [Feature Request] Display current working directory in CLI prompt
- [BUG] --channels ignored after v2.1.195 update (plugin:telegram@claude-plugins-official)
- Windows Terminal: Claude Code shows block cursor despite `cursorShape = "bar"`
- Worktree isolation silently disabled when lead CWD is stranded under .claude/worktrees/ (falls back to in-place edits)
- Worktree isolation inflates spawned-shell env past ARG_MAX → E2BIG on all Bash, unrecoverable without restart
- [BUG] Skill tool re-invoked after slash command already loaded skill
- Feature request: FleetView (`claude agents`) should show the repo/project per session row
- Background sessions dispatched from a non-git directory can never be deleted from the agents view: "worktree could not be removed (no git root for the worktree)"
- [DOCS] Bedrock docs omit `aws configure export-credentials` JSON shape support for `awsCredentialExport`
- Permission auto-approve should separate read/write/delete operations
- [FEATURE] Opt-in context inheritance for Agent-spawned subagents (inherit_context flag)
- [BUG] Support HTTPS proxy (https:// scheme) for HTTPS_PROXY environment variable
- /reload-skills does not refresh the slash-command autocomplete menu
- Feature request: session-only model override without changing global default
- [FEATURE] Exempt CLAUDE.md from compaction
- Custom model via ANTHROPIC_BASE_URL: no way to declare a context window >200k (status line & auto-compaction assume 200k)
- [BUG] [area:tool-use] [platform:macos] Write tool's full-file-replacement default causes irrecoverable data loss on governed, untracked state files — no append-only or protected-path mechanism available
- Feature Request: Support XDG Base Directory Specification for config/data directories
- [DOCS] [Fullscreen] New `wheelScrollAccelerationEnabled` setting added in v2.1.174 is undocumented in `fullscreen.md` and the settings table
- [Feature Request] Add per-subagent effort/thinking control to align with Agent SDK
- [Feature Request] Add session-level effort override without persisting to default configuration
- Mouse capture in TUI breaks Linux PRIMARY-selection middle-click paste (no opt-out)
- /exit warns about running background agents when none are running; agent sidecars never record completion
- [FEATURE] Session URL appended to commit messages and PR descriptions by default — should be opt-in
- [DOCS] `API_FORCE_IDLE_TIMEOUT` opt-out for the Vertex/Foundry 5-minute idle timeout is undocumented
- [DOCS] Errors reference omits the "Your organization has disabled API key authentication" message and v2.1.169 guidance about which credential source to check
- [BUG] plugin uninstall resolves by bare name, can silently uninstall an unrelated enabled plugin sharing the same name in a different marketplace
- [DOCS] `TaskCreate` tool reference omits v2.1.169 input-repair and unloaded-tool schema-in-error changes
- [DOCS] [Dynamic workflows] `/workflows` command docs don't state it opens immediately even while a turn is in progress
- Background sub-agents spawned by a subagent stay "Running" forever after the parent completes; TaskStop can't reach them
- [DOCS] [Plugins] `.in_use` PID lock files and the daily stale-marker sweep are not documented in the plugin reference
- [DOCS] Plugins reference does not document MCPB plugin cache invalidation and re-extraction behavior on Windows
- [DOCS] Background agent docs do not mention the pre-warmed worker dispatch path or its `env` settings behavior
- [DOCS] Agent view attach/detach docs omit the WSL + Windows Terminal stale/garbled frame that was fixed in v2.1.169
- [DOCS] Windows startup "Connect to GitHub" popup from Git Credential Manager is undocumented for background git commands
- Agent subagent results never delivered to parent in background job sessions
- [DOCS] Remote Control troubleshooting missing "stuck on reconnecting" entry for OAuth token refresh race during session resume
- [DOCS] Managed MCP policy (`allowedMcpServers`/`deniedMcpServers`) enforcement windows are not documented
- [DOCS] [Interactive mode] Up/Down arrow behavior on wrapped rows of a long input line is undocumented
- [Bug] TUI doesn't restore terminal state on SIGTSTP; suspend + paste wedges terminal
- [FEATURE] Let me mark an agent session as completed / dismiss it from the agents view
- AskUserQuestion: "Other" text field content remains after submission
- [BUG] bgIsolation containment check isn't realpath-canonicalized; symlinked cwd bypasses it
- [FEATURE] Make Workflow scripts (.claude/workflows/*.js) a distributable plugin component
- /resume can't find sessions started in a different directory (picker is cwd-scoped)
- [DOCS] Managed MCP allowlist and denylist docs omit `${VAR}` reference matching behavior
- [BUG] Skill activation emits a second `isMeta` message that orphans the Skill `tool_use` → 400 invalid_request (surfaced as "tool use concurrency")
- [DOCS] SendMessage docs omit relayed permission authority and auto mode blocking behavior
- [BUG] `claude rc` on the home directory: accepting the trust dialog never persists (`hasTrustDialogAccepted` stays false), trust loops forever
- Local transcript deletion (30-day TTL) should be configurable independently of data sharing
- Link Claude Code sessions to CLAUDE.md for cross-device sync
- [FEATURE] Voice input support for hands-free prompting in Claude Code CLI
- Pressing ← right after Enter (same input burst) silently drops the just-submitted prompt — the session idle-forks and no transcript is written
- [FEATURE] VS Code integrated browser support
- Feature: Real-time steering — send message mid-generation without queueing
- [FEATURE] Decouple LSP diagnostic auto-push from the LSP tool (keep navigation, drop publishDiagnostics)
- Bash tool stdout duplicated N times in tool results (N grows per call); command runs once — capture/render bug, not the shell
- [DOCS] Background session docs omit worktree-isolation behavior for spawned subagents
- [DOCS] Agent view dispatch input docs incorrectly imply `/logout` dispatches as a prompt
- [DOCS] Background session docs omit `$CLAUDE_JOB_DIR` temp-file behavior
- [DOCS] Agent view docs omit macOS Privacy & Security identity for background agents
- Add option to hide email address from welcome banner
- [Bug] Idle session busy-spins CPU with sched_yield in worker threads instead of blocking
- [Feature Request] Implement automatic context optimization and token usage tracking
- Feedback zur Claude Code VS Code Erweiterung (v2.1.226)
- vim editor mode: pasted line breaks become literal 'j' when paste isn't bracketed
- [Feature Request] Security-hardened code generation should not trigger model fallback
- [Feature Request] Add local environment trust mode to bypass security classifier for local-only operations
- No file path autocomplete suggestions for /cd in terminal CLI
- Session picker (claude agents) takes minutes to open sessions with large JSONL transcripts; agents view pegs CPU
- Bash permissions.deny rules not enforced in subagent (Agent/Task tool) shells — denied command executes, only post-hoc security warning
- PermissionRequest hooks: subagent prompts don't invoke the hook; when it did fire, a returned allow was ignored
- [FEATURE] Make AskUserQuestion tool timeout configurable
- Feature Request: Support global ~/.agents/AGENTS.md along with project-level AGENTS.md
- AskUserQuestion (multiSelect): user's typed "Other" free-text is silently dropped — model receives only the literal label
- [FEATURE] multi-agent IPC
- [FEATURE] Setting to disable copy-on-selection in the agents view (TUI)
- [FEATURE] Feature request: in-session /remote-control toggle to enable Remote Control on an already-running session
- Multiline input: arrow keys should navigate lines, not history
- [BUG] CLI does not persist conversation to local .jsonl in 2.1.113/2.1.114/2.1.227; /resume shows empty; cloud history also missing
- [DOCS] Agent view docs omit voice dictation support in the peek-panel reply input
- [FEATURE] Setting to suppress Ultraplan prompt in plan mode
- [DOCS] Agent view docs omit recovery guidance for an unresponsive background service
- [DOCS] Background side-query docs omit main-model fallback for gateways and Bedrock Mantle
- [FEATURE] Agent view: opt-out of worktree isolation and reuse heavy dependency directories
- [DOCS] Background-session docs omit macOS Full Disk Access behavior for protected home folders
- [DOCS] `/goal` docs do not explain evaluator waits for background shells or delegated subagents
- [DOCS] Background sessions docs omit effort-level persistence across idle wake
- Feature Request: Skill-scoped permissions
- [DOCS] Agent view omits 5-minute retirement for empty idle background sessions
- [DOCS] Plugin docs missing `CLAUDE_CODE_PLUGIN_PREFER_HTTPS` for GitHub sources
- [Bug] Assistant text blocks between tool calls not rendered in terminal output
- [FEATURE] Cowork: allow configuring the storage location for scheduled tasks (currently hardcoded to ~/Documents/Claude/Scheduled/)
- [DOCS] Permission docs omit PowerShell auto-approval behavior
- Routines in Claude Code can't use Claude.ai Connectors (e.g., Composio, Gmail, LinkedIn)
- [DOCS] WebFetch docs missing large-HTML truncation behavior
- [FEATURE] Option to disable Launch preview panel auto-open on Write/Edit
- [FEATURE] Enable marketplace auto-update through settings.json
- Feature Request: Message queue mode — queue messages instead of interrupting active tasks
- Add /resume (session resume) to desktop app
- [DOCS] Remote Control docs omit that claude.ai renames persist to the local CLI session title
- [DOCS] Headless and Agent SDK docs omit session auto-title background request behavior
- Bash tool grep()/find() shims: bare exec when BASHPID != $$ replaces the executing subshell — compound commands/loops silently truncate
- [FEATURE] Allow resuming Claude Code sessions across devices
- [FEATURE] Preview should allow localhost subdomains (e.g. playground.localhost)
- Feature request: Cancel/interrupt button for Remote Control (mobile)
- [PROPOSAL] Expose compact/session lifecycle hooks for external memory layers
- Programmatic session naming — hook-callable rename + Skill tool as hook target
- Feature: configurable reasoning effort level for subagents
- [FEATURE] Native OTLP Trace/Span Export for Claude Code
- Add persistent setting to always show thinking trace
- [FEATURE] Enable Auto mode support for Amazon Bedrock models
- [FEATURE] Allow configuring or disabling the URL Handler app install location (currently hardcoded to ~/Applications/)
- Better session names in /resume
- Stop hook re-fires despite stop_hook_active: true, causing /goal loop to spin after completion
- Background agents drop queued SendMessages and skip completion notifications
- Auto mode should be available on Pro/Max plans (not just Team/Enterprise)
- [BUG] Session .jsonl files pruned to 1-per-project despite cleanupPeriodDays: 9999 — correlates with autoUploadSessions: true
- [FEATURE] --bare mode: option to preserve OAuth/keychain auth while skipping context
- [FEATURE] Option to override ~/.claude/settings.json
- Add remoteControlEnabled setting to auto-start /remote-control on session start
- Support /btw command in VS Code extension
- feat: terminal bell (BEL) when Claude Code is waiting for tool approval
- Feature request: Session persistence & reconnection for Remote Control server mode
- [FEATURE] `/new` command to end current session and start a new named session
- Show extended thinking in CLI output (collapsed by default, toggle to expand)
- [FEATURE] Skill invocation tracking and usage analytics
- [FEATURE] The ability to queue up messages.via asking the user for further info on user input.
- Feature Request: Persistent Memory Across Context Compactions (59 compactions, built our own)
- Feature: Allow programmatic session rename from skills/commands
- Feature request: auto-check/update via winget before launch on Windows
- [Feature Request] Agent view: Add directory display and directory-based grouping for sessions
- Allow persisting thinking effort level across sessions
- Effort level 'max' should be persistable in settings.json for Max plan subscribers
- Add 'system' theme option to follow OS appearance
- Auto-rename sessions based on conversation context
- [FEATURE] Permission system should evaluate piped commands independently, not as a single string
- Voice: Add language configuration for speech-to-text transcription
- [FEATURE] Bring back thinking
- [FEATURE] VSCode: Support adding multiple code selections to chat context incrementally via shortcut
- Feature: combine remote-control with --dangerously-skip-permissions
- Add quiet/silent option for async hooks
- Voice mode: add toggle/continuous listening mode alongside push-to-talk
- Stream extended thinking output in real-time during interactive mode
- EnterWorktree: a stale/empty worktree directory silently resolves to the parent repo, so "isolated" commits land on the parent's checked-out branch
- [Feature Request] Real-time steering: priority message channel for redirecting Claude mid-execution
- Co-Authored-By default is inappropriate — stop branding user work
- Feature Request: Enable Remote Control for Team plans
- [FEATURE] Remote Control: Push notifications on iOS when permission approval is needed
- Feature request: Allow Claude to programmatically rename sessions
- Remote Control: add permission mode cycling from mobile app
- [FEATURE] Support `url` field in `launch.json` for preview feature
- Enable Remote Control for Claude Code sessions in Claude Desktop App
- [FEATURE] Worktree creation should sync from remote default branch, not local
- Feature: Push notifications for completed tasks in remote-control mode
- Allow resuming conversations from different directories
- Slash commands not supported in /remote-control UI
- [FEATURE] Allow slash command usage over remote-control
- [FEATURE] Configurable auto-memory storage location (memoryDirectory setting)
- Allow scrolling back to view full conversation history in CLI TUI
- [FEATURE] Allow relocating active sessions to a new working directory
- [FEATURE] Add PostWorktreeCreate hook (or setup command) for environment initialization in worktrees
- [Feature Request] Modern text input: click-to-position, text selection, and standard editing in the prompt
- [BUG] Fullscreen TUI: footer navigation (Left/Right arrow) re-enables mouse capture, ignoring CLAUDE_CODE_DISABLE_MOUSE=1
- Feature Request: Named account profiles for quick switching between multiple claude.ai accounts
- [iPadOS] App crashes when opening any session in Code tab (v1.260618.0) — main-thread stack overflow in SwiftUI
- Findings-list renderer wraps long file paths mid-word, causing overlapping/garbled text
- [FEATURE] Preview: Allow configurable external URL whitelist for OAuth and other third-party required flows
- [BUG] Experiência ruim com sequência de ações não feitas ao mesmo tempo
- [FEATURE] Paste images from clipboard directly into Claude Code (Windows)
- Feature Request: System notifications when Claude needs attention or completes tasks
- Add environment variable to configure .claude config directory location
- Feature Request: Option to control auto-scroll behavior when submitting questions
- [FEATURE] "Add a VS Code setting to disable auto-opening diff tabs when Claude edits files."
- [FEATURE] Support for multiple accounts / profiles
- Inter-session communication for multi-Claude workflows
- [FEATURE] ~/.claude.json should be stored in ~/.claude
- [FEATURE] Allow custom .claude/agents/ definitions as agent team teammates
- [FEATURE] Bring 1M context window to Max plan subscribers
- [FEATURE] Add Max thinking option for Claude Opus 4.6
- Feature: Model selection in skill frontmatter
- [Bug Report] Spinner Status Words feel unprofessional and dismissive
- [FEATURE] Trusted workspace patterns to skip trust prompt for git worktrees
- [FEATURE] PermissionRequest and PermissionResponse hook events for metrics
- [FEATURE] Add settings to hide token counter and version display in status line
- Feature Request: Meaningful plan file names
- [FEATURE] Provide way to exclude parent CLAUDE.md files from auto-loading in subdirectories
- Feature Request: Multi-Account Profile Support
- New 'native' installer: Bun/Zig hype train delivers 2010 anti-patterns: no version control, rollbacks, or professional package management
- [FEATURE] Support terminal notifications when running inside tmux
- [FEATURE] Auto compact and pre compact param separation
- Bundled ugrep (Bash `grep` shim) balloons to 14–16.6 GB RSS on bounded-repeat -o patterns that GNU grep handles in 3 MB — repeated OOM kills on a production box
- [BUG] Claude should not hijack the mac-os EDITOR env variable.
- [BUG] Feature Request: PreToolUse Hook Context Auto-Injection API
- [Feature Request] Allow configuring external editor for Ctrl-G prompt editing
- [FEATURE] Allow some commands to execute asynchronously
- Feature Request: Add 'Allow always (save to settings)' permission option
- [FEATURE] Tool result transform hook for content sanitization
- [FEATURE] Add setting to configure default option in plan mode approval dialog
- [FEATURE] VSCode Extension: Display context usage percentage in UI
- [FEATURE] Recursive skill discovery - scan subdirectories in ~/.claude/skills/
- [FEATURE] Claude Code Mobile Companion App
- [FEATURE] Terminal Title Configuration (Script-Based, like statusLine)
- [Feature Request] MCP Context Isolation - Assign MCPs to Forked Agent/Skill Contexts
- [FEATURE] Persistent in-place plugin loading (no cache)
- Feature Request: Add PreCompact and PostCompact hooks for custom context management
- [FEATURE] Chat only toggle in Code session
- [Feature Request] Move agentic instructions from CLAUDE.md to vendor-agnostic AGENTS.md
- Feature Request: Add Dart/Flutter LSP support
- [FEATURE] Allow providing tool_result via stdin when resuming session with pending tool_use
- Feature: Parse compound Bash commands and match each component against permissions
- [BUG] There is no indication to the user when Claude sets `dangerouslyDisableSandbox: true`
- [BUG] Switching TUI renderer (/tui) silently discards a just-performed /rewind
- permissions.deny Read rules cannot fence the project auto-memory directory (~/.claude/projects/<cwd-slug>/memory/*)
- [Feature Request] Add configurable editor option for Ctrl+G shortcut
- [FEATURE] Add support for AWS Bedrock service_tier parameter (flex, priority, reserved, default)
- [FEATURE] Allow for updated input in `PermissionRequest:ExitPlanTool:Approve` hook
- Mobile Companion App for Remote Session Interaction
- Feature Request: Add YOLO mode (bypassPermissions) to Shift+Tab permission mode cycle
- [FEATURE] support mouse scrolling in tmux
- Session state should persist across git worktrees
- [FEATURE] Automatic Model Switching for Plan Mode
- Enable LSP tool to connect to installed language servers
- [FEATURE] Support `ref` and `path` parameters in plugin source schema
- [FEATURE] Auto-cleanup Chrome tab groups when Claude in Chrome sessions end
- Session-env snapshot file breaks when sourced: unquoted LS_COLORS/SSH_CONNECTION/LESSCLOSE spam 'command not found' on every Bash call
- Session becomes unreachable ("No conversation found") after git worktree remove deletes its own working directory
- [FEATURE] Support targeting specific Chrome instances/profiles for Claude in Chrome integration
- [BUG] /context uses 200k window for claude-fable-5, which has a 1M context window
- [FEATURE] Add browser selection option for Chrome extension (support non-default browsers)
- Bundled ugrep OOMs the host: -E with two bounded {0,N} intervals explodes DFA construction to multiple GB
- [Feature Request] Increase API image size limit from 5MB to 10MB to match claude.ai
- [FEATURE] 1. Agent Hierarchy in Hook Events, 2. Intermediate Text Output Hook, 3. SubagentStart Hook
- Claude in Chrome: Allow browser selection instead of opening default browser
- [FEATURE] enable extended thinking for subagents, currently not possible
- [FEATURE] PostCompact Hook Event and Compaction Content Control
- [FEATURE] GitHub CLI Support for iOS Claude Code
- [Feature Request] Include actual context window usage in statusline JSON
- [FEATURE] Reload local plugins on startup
- [BUG] Interactive session with custom CLAUDE_CONFIG_DIR writes no transcript JSONL anywhere (headless -p works) — 2.1.207+
- [FEATURE] Delete Claude Code sessions
- [FEATURE] Add hook for when Claude is waiting for user input
- [FEATURE] OSC 8 Hyperlinks for Clickable File Paths
- [Feature Request] Add option to skip trust prompt when .claude/settings.local.json allows it
- [Feature Request] Subagents should inherit parent agent context to avoid redundant codebase scanning
- Screenshot Support via Clipboard Paste in Claude Code CLI
- Feature Request: Session Branching / Conversation Forking
- [FEATURE] Expose /usage and /context data in statusLine JSON
- [Feature Request] Claude Desktop: Option to execute commands in WSL instead of Windows
- [FEATURE] Option to disable notifications
- Add runtime keyboard shortcuts for permission mode toggling
- [FEATURE] vscode extension: font family and size
- [FEATURE] .NET 9 or 10 SDK support in Claude Code for web runtime environment
- Expose token usage data to statusline scripts
- [FEATURE] /pr_comments ranks higher than /commit when typing /com - improve fuzzy matching, enter selects unexpected command and allow disabling built-in commands
- [FEATURE] Feature Request: Allow configuration to prevent auto-collapsing of long Bash outputs
- [FEATURE] Rename tabs in VS Code Extension
- [FEATURE] Persistent user-level MCP configurations (enable/disable)
- [FEATURE] Add native GitHub Issues integration for reading and implementing issues
- [FEATURE] Delete and rename chats in VS Code extension
- [FEATURE] Migrate checkpoint/rollback functionality to VSCode extension
- [FEATURE] Ability to enable or configure automatic updating of a marketplace and plugin
- [FEATURE] Add Skill Autocomplete to CLI (Parity with VS Code Extension)
- [FEATURE] Add support for subdirectories in skills
- Add hook for user input/question events (UserInputRequired)
- [FEATURE] Add deep research mode as a /research slash command
- [FEATURE] Allow Claude Code Web to start sessions from branches other than default branch
- [FEATURE] Support for Plugin Dependencies and Shared Resources
- [BUG] Fix ${CLAUDE_PLUGIN_ROOT} in command markdown OR support local project plugin installation
- [BUG] Streaming partial-JSON parser silently produces empty MCP tool arguments (accumulator shear)
- Enhancement: Add --quiet flag to suppress tool call output
- [Bug] Inherited stale CLAUDE_CODE_SESSION_ID causes silent transcript loss without warning
- [FEATURE] Add user-configurable keyboard shortcut actions
- [FEATURE] Disable or Customize User Input Background Highlighting
- [FEATURE] Add Option to Always Show Claude's Thinking
- SubagentStop hook cannot identify which specific subagent finished due to shared session IDs
- Feature Request: Restore Time & Token Indicators + Add Context Usage Display
- Feature request: Claude Code profiles with isolated memory, commands, hooks, and settings
- Agent Context Detention in Hook Events
- [FEATURE] Add --no-auto-compact command switch to Disable the auto-compaction
- Feature Request: Support AGENTS.md.
- Allow customization of mode switching shortcuts (Alt+M vs Shift+Tab)
- Feature Request: Add /reloadSettings command to reload configuration without restart
- Feature Request: Add /copy command to copy messages to clipboard
- [FEATURE] Ctrl+Enter for newlines conflicts with standard app conventions - request customizable shortcuts
- History accumulation in .claude.json causes performance issues and storage bloat
- Feature Request: Add model selection support for custom commands (similar to subagents)
- Allow viewing and editing content of “pasted text” blocks before submission
- Cannot work with git worktrees due to directory restriction
- VSCode extension: hotkeys for navigation
- Insert a new line with Enter key instead of sending a message
- Support shift+enter in most terminals
- Usage Metrics Visibility for Max Subscribers
- Simplify copying of markdown output
- [FEATURE] Richer permission-prompt notifications — sharing a workaround (ref #32952, #29827)
- Bash sandbox (bubblewrap) corrupts `!` to `\!` in commands, making the sandbox unusable for agentic workflows
- [SECURITY-BUG ] Permission prompt is silently replaced by a newer pending approval (LIFO stack), allowing the wrong command to be approved in manual-approve mode
- v2.1.173 regression: TUI writes no transcript records when CLAUDE_CODE_CHILD_SESSION is inherited from a parent claude session — silent data loss, --resume broken
- Feature Request: Global memory / cross-project shared memory
- [BUG] Opening session sidebar wipes API key auth for current session
- [FEATURE] Feature request: quick "jump to first message" button in long conversations
- [BUG] Agent view (claude agents) mojibakes pasted CJK input (UTF-8 read as Latin-1, C1 bytes dropped) and ignores tui:default / CLAUDE_CODE_NO_FLICKER=0 — Linux, v2.1.173
- [FEATURE] ADD DELETE BRANCH FONCTIONNALITY
- [FEATURE] Show current model name in the CLI status bar
- [BUG] /goal Stop hook emits markdown to stdout, fails JSON schema validation ("JSON validation failed")
- Background tasks (run_in_background Bash + Agent) stay "Running" in the panel forever after completing
- Re-Open FleetView Manually with a /FleetView command!
- Feature request: configurable retention period for claude.ai conversation history
- Agent tool ignores `effort:` frontmatter in subagent .md files — inherits global effortLevel instead
- 【@@📞📞📞𝐎𝐟𝐟𝐢𝐜𝐢𝐚𝐥🧲Przewodnik po LOT @📞📞📞POL@SKA! 𝐓𝐞𝐥𝐞𝐟𝐨𝐧】 🧲Jak skontaktować się z centrum obsługi LOT?📞📞📞
- [BUG]
- [BUG]
- [BUG]
- [Feature Request] Allow user model selection instead of forcing Opus 5
- [FEATURE] Agent-to-agent messaging with wake-on-delivery — every multi-agent setup is hand-rolling a message bus on the filesystem, and the sprawl IS the queue
- [Feature Request] Add security scanning capability for repository vulnerability detection
- [Feature Request] Add model persistence option to prevent switching from user-selected model
- [Feature Request] Add security audit mode with safeguards flag for project analysis
- [FEATURE] NON-VAGUE Usage reset timers
- [Feature Request] Add true Push-To-Talk voice mode with configurable hotkeys
- [BUG] /rewind (Esc Esc) silently reverts/loses code — destructive "Restore code and conversation" is the default with no confirmation
- [Feature Request] Add direct feedback submission command (/bug) for bug reports
- [FEATURE] VS Code extension - Play sound when requires action
- sandbox.enabled: true triggers a synchronous full-tree directory walk on startup → multi-minute hang in large workspaces
- [BUG]
- [Bug] Allow rules under `~/.claude/` show as loaded per /permissions but don't match at runtime
- [Feature Request] Restore access to claude-opus-4 model variant
- [BUG] In-session /resume picker collapses to ~0 rows inside a git worktree and never recovers (Ctrl+W restores; `claude -r` from the shell is unaffected)
- [BUG] Highest-priority saved instruction ("respond in text before any tool call") repeatedly broken within same session
- Stop hook decision:block is rendered as 'Stop hook error', making intentional instructional blocks look like failures
- [Feature Request] Add verbosity/logging options to keep user informed during execution
- Agent tool: isolation:"worktree" subagent read parent session's private transcript outside its worktree
- Background sessions (claude agents / fleet) don't honor permissions.defaultMode: bypassPermissions
- plugin marketplace update fetches new version but doesn't update installed_plugins.json
- [Bug] Even on cyber verification program, Fable won't perform BASIC defensive security work
- [BUG] sandbox.filesystem.denyWithinAllow not enforced for Edit/Write tools
- Vim mode: Ctrl-[ dismisses AskUserQuestion prompts instead of entering normal mode
- [Bug] Grep and Glob tools not registered in Claude Code 2.1.154+
- [BUG] malformed tool_reference content block injected into history breaks vLLM chat-template rendering (jinja2 TemplateError: "Unexpected item type in content")
- [FEATURE] options to "ignore" or white list MCP and other tools, so this session ignores anything not specified as an external needed tool
- skillOverrides silently ignored for plugin-provided skills — runtime hard-codes source==="plugin" to "on" while the /skills UI implies overrides work
- [BUG] Contradictory guidance on Write() vs Edit() permission rule matching in settings.json
- Trust dialog reappears every session — `hasTrustDialogAccepted` not persisted on `/exit`
- [FEATURE] Search the sessions list in claude agents
- [BUG] Agent tool model parameter accepts only family aliases; aliases resolve to the latest model and cause a 401 for teams with a restricted allowlist
- .claudeignore not enforced by Bash or Edit tools — only blocks Read/Grep/Glob
- [BUG] @file reference bypasses PreToolUse hooks (e.g. .env read block)
- [FEATURE] paste screenshot without text
- Bash permission prompts fire inconsistently even for commands matching permissions.allow / defaultMode: "dontAsk"
- MCP server instructions are silently truncated when multiple servers are configured
- Worktree isolation (Agent/Workflow) forks from the default branch, not the current branch — wastes agent fan-outs on a divergent base
- [BUG] [-p / print mode] AskUserQuestion tool was available in 2.1.185 but removed in later versions — please make it opt-in instead of hard-disabling
- [BUG] copyOnSelect: false is ignored by the fullscreen renderer (honored by the default renderer) — v2.1.206
- [BUG] possible?Prompt injection? detected in a Claude Code skill's tool output (dataviz skill)
- Agent(run_in_background: false) does not block — returns spawn metadata instead of the agent's result (v2.1.227)
- [FEATURE] Feature Request: Ability to reload/refresh skill definitions during a session
- [BUG] Workspace sandbox config silently dropped for sessions/subagents rooted in nested project directories (sandbox escape)
- [Feature Request] Add option to redisplay "How's Claude doing?" prompt after dismissal
- [FEATURE] Fullscreen TUI Ctrl+R only searches the last 100 prompts — restore classic renderer's full-history exact search
- [FEATURE] how to switch claude code account in vscode its needed
- [FEATURE] USAGE BARS
- [FEATURE] multi-line command prompt in cli interface (for visual studio)
- [FEATURE]
- Feature request: compact/minimal-output mode to suppress inline tool-call blocks
- [Feature Request] Add token usage audit reports (session, 5-hour window, weekly) with leak detection for cross-agent skill transfers
- [BUG] a /goal Stop hook cannot be cleared from inside the session
- Feature Request: Informational Exit Code for Hook System (exit code 3 / `display: "info"`)
- [FEATURE] User configurable plan presentation timeout
- [FEATURE]
- [Feature Request] Option to disable alternative verbs in progress messages
- [Feature Request] Allow Claude Code agent usage in isolated educational environments
- [Feature Request] CVP eligibility review for authorized security assessment requests
- [Feature Request] Add allowlist/exemption mechanism for defensive security workflows
- [Feature Request] Add manual trigger command for plan operation menu (e.g., `/plan start`)
- [Feature Request] Support for Cloudflare Zero Trust authentication headers
- [FEATURE] Claude Code VS Code Extension - disable the spinner icon
- Monorepo: no way to see which working directory a session is in (especially on mobile)
- [Feature Request] Add direction parameter to /compact command for contextual summarization
- [FEATURE] Discover and list models from custom base URL in `/model` picker
- [FEATURE] VSCode extension doesn't show which account is logged in
- [Feature Request] Allow whitelisting own projects for security audits without model downgrade
- [FEATURE] Make Skills created on Claude.ai (web) available in Claude Code via slash commands
- Claude Browser Tab Realy
- [FEATURE] empty state prompt suggestion
- Feature request: /permissions should support a global scope
- [BUG] Daemon bg/spare sessions leave title-only .jsonl stubs in the resume picker; resuming errors, then starts a blank session that re-submits the last prompt
- [BUG] `/code-review` tells its background agent to call `ReportFindings`, which the background agent doesn't have
- Feature request: default the transcript view to show extended thinking expanded
- [FEATURE] Disable individual claude.ai connectors at user scope, not just per-project
- [FEATURE] Auto-mode excluding editing files
- [Feature Request] Auto mode should respect allow list settings from non-auto mode
- [Feature Request] Add model selection lock to prevent automatic switching from Fable to Opus
- Show local timezone (or allow configuring it) in /usage output
- [FEATURE] CLAUDE_CODE_OUTPUT_STYLE env var to override outputStyle for headless runs (claude -p)
- [Feature Request] Display current working directory/project context in CLI interface
- [FEATURE] "Hallucinating..." funny string
- [Feature Request] Add persistent status line displaying active model
- PreToolUse if:-gated agent hooks fire nondeterministically on non-matching Bash commands, and agent ALLOW verdicts sometimes convert to denies
- [FEATURE] Tell the model its own context-window usage, the way it's already told the date
- [BUG] 2.1.211 "always allow rules save at repository root" breaks unattended runs in a linked git worktree under a worktree-scoped filesystem sandbox
- [BUG] ConfigChange hook emitting blocking decision is not respected by claude code CLI
- [Feature Request] Support direct clipboard image pasting from screenshot tools
- [Feature Request] Add option to disable mouse input and use keyboard navigation only
- [Feature Request] Ask User Question Tool: Support sending notes-only response with '0' key
- [FEATURE] Setting to permanently disable in-product feedback/survey prompts
- [FEATURE] Incognito mode for code review
- [Feature Request] Option to use git branch name as the session title (desktop app has no way to automate this)
- Remote control of a Claude Code session requires being physically at the host machine to enable
- Update terminal tab title to show current/last working directory
- allowed-tools does not pre-approve Read/Write for paths in the user's home directory (only ${CLAUDE_PLUGIN_ROOT} paths work)
- [Feature Request] Model picker lacks session-only switching option
- Allow disabling individual built-in skills to reduce fixed per-session context overhead
- [FEATURE] Ctrl+C destroys the draft prompt with no recovery - preserve cleared input in history like double-Esc does
- [Feature Request] Reduce frequency of feedback prompts and add option to defer
- Configuración nativa de idioma en vez de depender de instrucciones en CLAUDE.md
- [FEATURE] Render markdown checkboxes in chat
- External-edit attachment misattributes agent's own Bash-tool file writes, wraps them with a "don't tell the user" instruction
- Feature request: disable/replace whimsical status-spinner text with professional alternatives
- [FEATURE] Screen Shot Hotkey Automation
- [FEATURE]
- [Feature Request] Reduce verbosity of CLI output and add concise mode option
- Workflow tool eager-loads ~4k tokens despite being opt-in/explicit-trigger only — defer via ToolSearch or add a toggle
- Warn or confirm when a new session reuses the --name of another active session
- Emit OSC 8 hyperlinks for displayed file paths (like URLs)
- Auto-mode classifier too restrictive for private single-operator homelabs
- [Feature Request] Add option to disable response placeholder/prefix
- [FEATURE] Remote access to sessions should be the default, not opt-in per session
- [FEATURE] Persistent context usage indicator (not just a warning near the limit)
- Feature request: Korean (한글) bilingual display for CLI chrome text (spinner verbs, "Tip:" lines)
- /add-dir with no argument should list the currently included directories
- [DOCS] MCP Tool Search behavior conflicts between docs.claude.com and Anthropic partner training (Claude Code 101)
- Bundled workflows (code-review, deep-research) offer no model override — all 10-15 subagents inherit the session model
- [DOCS] Skills reference still implies `/verify` and `/code-review` can run automatically
- [FEATURE] Display model in status bar
- [DOCS] Hook if-condition documentation omits the changed scope of single-segment dir/** patterns
- [DOCS] Hooks reference omits the SessionStart “fork” source
- [DOCS] Ultrareview documentation omits the no-merge-base fallback
- [DOCS] subagentStatusLine documentation omits the new reasoning-effort payload field
- [DOCS] Tools reference is missing the EndConversation tool
- [DOCS] Permission docs do not distinguish dir/** scope for allow versus deny and ask rules
- [Feature Request] Remove or make optional API safeguards and Opus fallback behavior
- Expose remaining token count as env var for statusLine
- [FEATURE]
- [Feature Request] Disable mouse interactions and implement keyboard-based text selection in TUI
- [Feature Request] Add `--advisor` flag to Claude Code CLI
- [DOCS] Enterprise authentication docs omit the login surfaces enforced by `forceLoginMethod`
- [DOCS] Prompt-caching docs do not explain mid-conversation system-block caching through gateways
- [DOCS] Ultrareview docs list only a bare PR number, not accepted PR reference formats
- [DOCS] Agent view docs omit the `/resume` picker for sessions removed from the list
- [DOCS] MCP auto-background threshold and disable setting are missing from the docs
- [DOCS] Environment-variable reference missing the per-session subagent spawn cap
- [DOCS] Environment-variable reference missing the WebSearch per-session cap
- [DOCS] Auto mode configuration guide missing `claude auto-mode reset`
- [DOCS] Document `/subtask` and the new `/fork` background-session behavior
- Show live session/weekly usage % and reset countdown in CLI (e.g. /usage)
- Max plan models (Opus 4.8, Fable) not available in Claude Code CLI
- Markdown links in responses are not clickable in terminal (Ghostty) — need raw URL fallback or OSC 8 hyperlinks
- Suggested replies disappear on first keystroke — no way to send them as-is without retyping
- [FEATURE] Enable claude code desktop to start multiple new chats from remote
- [FEATURE] Remove stupid autocomplete suggestions in Claude Code.
- [Feature Request] Add configurable concurrency limiter for parallel agents with lower default
- Skills and commands share the /name invocation space with no distinction
- MCP tool namespaces are opaque and inconsistent — cannot tell which server a tool belongs to
- [DOCS] Claude in Chrome docs describe `save_to_disk` approval only, not disk write and returned path
- [DOCS] Interactive mode Vim reference missing NORMAL-mode `s` and `S` substitute commands
- Solicitud de funcionalidad: localizar los verbos del spinner y los tips rotativos (soporte i18n/locale)
- Agent behavior
- [DOCS] Permissions docs omit repository-root storage for always-allow rules in git worktrees
- [DOCS] `--setting-sources` docs do not state that omitting `project` skips nested `.claude/rules/*.md`
- [DOCS] Sub-agents resume section omits that an explicit model selection persists across resume and follow-up
- [DOCS] Hooks guide does not document PreToolUse `ask` flooring auto mode for unsandboxed Bash
- [DOCS] Headless CLI docs omit `--forward-subagent-text` and `CLAUDE_CODE_FORWARD_SUBAGENT_TEXT` for stream-json subagent output
- [FEATURE] native session-to-session (agent-to-agent) communication
- Setting to hide built-in vim mode indicator when custom statusLine renders it
- [Feature Request] Persist custom chat name across session clear operations
- CLAUDE.md @ import always inlines content — need a "reference only" mode for lazy-loaded rules
- [DOCS] Screen reader accessibility docs still say permission mode changes are not announced
- [DOCS] Auto memory docs omit the error when `MEMORY.md` exceeds its read limit
- [DOCS] Auto mode classifier model selection for external sessions is undocumented
- [DOCS] Bash and PowerShell tool docs omit timeout-triggered auto-backgrounding messages
- [DOCS] Agent tool security guidance omits indirect prompt injection from subagent-read content
- [DOCS] Sandbox docs omit protection for late-appearing `.claude/*` symlinks
- [DOCS] Agent SDK MCP docs omit immediate connection behavior for initialize-registered servers
- [DOCS] Agent view docs omit task checklist persistence after `←` navigation
- [DOCS] `claude agents --effort ultracode` behavior for dispatched sessions is undocumented
- [DOCS] Skills and custom command docs omit unmatched positional placeholder behavior
- [DOCS] Grep content mode does not document pagination and past-end results
- [DOCS] Plan approval docs do not explain untouched-plan handling or stale snapshot protection
- [DOCS] Plugin MCP lifecycle docs omit connection preservation during mid-session server resync
- [DOCS] Bash backgrounding docs do not explain that `cd` does not change the session directory
- [DOCS] Agent SDK hook callback timeout behavior is incomplete for unattended sessions
- [BUG] PermissionRequest hook stops firing after CLI restart on Windows (v2.1.233)
- [DOCS] Dynamic workflows docs do not explain that the `ultracode` keyword requires human-originated input
- [DOCS] Headless stream-json control protocol does not document `set_model` validation or error responses
- [DOCS] Headless stream-json input does not document blank-line handling for Windows SDK hosts
- [DOCS] Glob tool docs omit working-directory null-byte errors
- [DOCS] Environment variable reference does not document scientific notation for numeric limits
- Feature Request: PreToolUse hook support for deny/block
- [FEATURE] No way to insert a newline in the agent view dispatch input (Shift+Enter dispatches)
- Auto mode: AskUserQuestion timeout (60s) should be configurable, or the session should wait instead of auto-selecting
- [Feature Request] Implement cache-aware CLAUDE.md editing to prevent mid-session invalidation
- /model: add session-only flag to avoid overwriting saved default
- [FEATURE] distinguishing user prompts from assistant responses in long threads
- [BUG] Missing session closure option when reaching a limit
- Native image paste support in Claude Code CLI/TUI using data:image/png;base64
- [Feature Request] Display current model name below input area
- Suporte a outros idiomas (Português) no microfone de ditaé embutido
- [BUG] Resuming a session on a later calendar day silently invalidates the entire message-history cache (currentDate block re-synthesized on resume)
- [FEATURE] Voice input (microphone) does not support Russian / non-English speech recognition language selection
- [Feature Request] Subagents spawned via Agent tool should inherit CLAUDE.md
- Built-in guard against Claude reading/printing credential-file contents (.env, shared.env, etc.)
- Allow customizing the thinking/status spinner messages
- [FEATURE] Interactive incognito/ephemeral session — extend --no-session-persistence beyond --print
- [FEATURE] PreExitPlanMode hook — intercept before ExitPlanMode call for planning artifact adversarial audit invocation
- [DOCS] Plugin marketplace "Owner fields" table omits schema-defined owner.url
- Allow customizing the "working" spinner verbs (and ideally sync across surfaces)
- Let me expand & edit a collapsed multi-line paste in the prompt (in-place, no external editor)
- Support multiple named profiles for `claude auth`
- [DOCS] Alpine Linux setup docs missing bash dependency — install.sh fails with not found
- [Feature Request] Add read-only mode for code analysis without modifications
- [DOCS] Plugin reference does not document that project `pluginConfigs` are ignored
- [DOCS] Amazon Bedrock authentication docs omit the Windows AWS credential-resolution stall guard
- [DOCS] Installation docs do not explain externally managed launcher protection
- [FEATURE] Bring Remote Control to the Desktop app's native Code sessions (parity with CLI)
- [FEATURE] Sticky Prompt with Scrollable Conversation
- [BUG] autoMemoryEnabled=false does not suppress the ~11-16k memory preamble (re: closed #44829)
- Desktop app: helper-tool install prompt doesn't support Touch ID
- [BUG] Mobile app (Android): Claude Code code blocks render in a proportional font, not monospace
- [BUG] M365 Connector reverts to read only when *any* scheduled task runs.
- [Bug][cyber] Building drone flight UI with live video + telemetry connection-status HUD blocked (req_011CcYJgUdojjgA9CLvAaxrz)
- [Bug] Agent execution loop causes corrupted state
- [Bug][cyber] Safety block halted routine GUI work on a drone telemetry/video ground-station HUD (req_011CcYJUCfHYwm1Zzc8uwssh)
- [BUG] Desktop SSH remote BinaryDeployment deploys linux-x64-musl to a glibc WSL2 host; fails "not runnable" / statx: symbol not found
- [Bug][cyber] Correcting a too-wide on-screen video aspect ratio via ffmpeg parameter tweaks (req_011CcYKHENYHwr9vFJzc6eDX)
- [Bug][cyber] Block halted debugging of a captured H.264 video stream decode/extraction pipeline (req_011CcYNMVAVYbm5pwJSrZdpJ)
- [BUG] Subagents going idle
- [Bug][cyber] False cyber block while building drone flight UI with live video, telemetry, and connection-statu (req_011CcYLSQEyq2CeAsBAgK4Pz)
- [Bug][cyber] Safety block prevents writing or reviewing code that reads drone telemetry sensor data (req_011CcYR3zvFDq6wKPwZEMFur)
- VS Code extension ignores sandbox settings.json / /sandbox unavailable (approval-fatigue fix unreachable in IDE)
- Windows: non-ASCII file paths corrupted (surrogate-escaped) when passed to PreToolUse Python hooks
- False positive: API safeguards block legitimate security code reviews via custom subagents [skip-publish-guard]
- [BUG] Desktop app: input draft loses markdown formatting after switching session tabs
- Image attachment behavior changed in v2.1.195: binary images now sent as file paths instead of inline
- [BUG] Windows Desktop app: `!` interactive shell forces PowerShell, ignoring defaultShell:"bash" (CLI works with identical settings)
- New chat submission steals focus to Chrome window (macOS Space switch) when Claude in Chrome is paired
- [BUG] Claude in Chrome MCP `navigate` prepends `https://` to `file://` URLs → chrome-error (same class as #19911)
- [Bug] Image references removed during JIRA description edits cause permanent screenshot loss
- [BUG] computer-use MCP: batch actions disrupt window Z-order across all monitors on Windows multi-monitor setup
- Auto-compact does not trigger after context exceeds 80%-100% (2.1.162)
- [Bug] Auto-compact window defaults to 500k tokens without user configuration
- Windows/Git Bash: Bash tool PATH becomes Windows/Unix hybrid after a plugin/MCP connects mid-session, breaking node/git/coreutils
- Desktop: "Add from a repository" (custom plugin marketplace) gone from "+" — now opens Directory only (1.15962.1)
- [BUG] Windows: chat input + push-to-talk mic freeze when SuperWhisper (dictation app) is running
- [Bug] Full Screen Mode: Auto-scroll behavior prevents manual scrolling
- [BUG] Chinese (CJK) characters displayed with orange highlight boxes in conversation/diff panel in vscode
- [Bug] Agent fails to parse tool-call error output, enters read-retry loop wasting thousands of tokens
- [BUG] Autonomous agent runs start without user consent and consume tokens (100+ times in June)
- [Bug] Anthropic API Error: Output blocked by content filtering policy on large Compose UI generation
- [MODEL] Refund policy does not account for model incompetence — multiple levels of failure
- Prompt injection committed as a user message via --replay-user-messages (VS Code extension), not reproducible in bare CLI
- [Bug] Intermittent file-read failures and silent cross-reads in workspace mount access layer
- Desktop app sidebar loses all session history after restart
- [BUG] Cowork Dispatch not responding on macOS
- Leading multi-slash prompts expand inconsistently (31/35 only the first, 4/35 all); UserPromptSubmit hooks can't see which were dispatched
- Worktree-isolated agent's worktree is cleaned at turn end while its background children still reference it
- Workflow script dialect and loader transform are undocumented — no offline parse-check or test entry point
- [BUG] Weekly usage counters not reset after scheduled weekly reset (Max 20x)
- [FEATURE] Enable Mic in Claude Code
- Workflow subagents receive the session's teammate roster + SendMessage — intended? (works well; docs + an off-switch would help)
- [FEATURE] Show the actual model of each spawned subagent in the task/progress UI
- Resuming a session can silently change the model (--resume should pin or warn)
- [FEATURE] Expose bypass permissions in the VS Code extension via a per-session, type-to-confirm gate
- [FEATURE] 채팅 제목 자동생성 언어를 한국어로 설정하고 싶다
- [DOCS] Dynamic workflow progress view omits left-arrow back navigation
- [DOCS] MCP configuration docs omit per-server `request_timeout_ms`
- [DOCS] Background-agent update docs still describe stale workers surviving until attach
- [DOCS] `EnterWorktree` docs omit confirmation for worktrees outside `.claude/worktrees/`
- [DOCS] `/doctor` docs omit checked-in `CLAUDE.md` trimming of codebase-derivable content
- [FEATURE] Expose built-in Grep/Glob tools in the VS Code native extension (parity with standalone CLI)
- VS Code extension: add a setting to collapse/hide tool-use cards by default
- [Feature Request] Add CLI option to set custom terminal tab title for Claude Code sessions
- claude desktop: detect and load every nested .claude / CLAUDE.md across subfolders
- [FEATURE] Stream Claude's thinking live in the CLI — the spinner says it's thinking; we'd like to know what
- [FEATURE] Option to COMPLETELY disable "accept edits" on
- Sonnet 5 selected and displayed, but every response is silently served by claude-fable-5, consuming capped Fable allowance
- [Feature Request] Add ability to clear queued prompts without interrupting execution
- Claude config path should be configurable (security/sandboxing blocker)
- [Feature Request] Add ad-hoc code signature to CLI binary for stable identity verification
- [FEATURE] show which model each background subagent is using in the tasks panel
- [Docs] Settings reference omits `switchModelsOnFlag`; `skipWorkflowUsageWarning` is app-written but unexplained
- [DOCS] MCP docs list only `workspace` as a reserved server name; v2.1.205 also reserves "Claude Browser" and "Claude Preview"
- [DOCS] `/checkup` alias for `/doctor` added in v2.1.205 is not documented in the commands reference
- [DOCS] Background task notifications are not documented as explicitly marking "no human input has occurred" so SDK consumers and Notification-hook handlers cannot rely on the v2.1.205 anti-fabrication guarantee
- [DOCS] Auto mode docs do not describe the v2.1.205 prompt fallback for `rm -rf` whose target variable cannot be resolved from the conversation
- [Feature Request] Add Quick Answer control to Claude Code CLI
- [Feature Request] Reduce verbose output and latency for simple commands like SQL execution
- Remote Control (/rc) re-enables itself / cannot be permanently disabled
- [Feature Request] Add session-scoped model selection without persisting global default
- [Feature Request] Reduce verbosity in CLI output and improve conciseness
- [FEATURE] Let a /code-review (or reviewer agent) session return a resumable ID for targeted re-review after fixes land
- Disable 'remember last model selection' for new windows
- [Bug] Claude model upgrade not prompted before consuming additional tokens
- [DOCS] [Agent view] `Needs input` state legend and version history omit the v2.1.203 fix for stale state after the question is answered
- [DOCS] `claude agents` crashes opening a stopped session whose conversation is already open in another session; agent view version history stops at v2.1.202
- [DOCS] [Agent view] Dispatch composer in `claude agents` is documented only for the unavailable-slash-command hint; v2.1.203 fix to preserve the typed message is not described
- [DOCS] Background daemon auto-upgrade failure now preserves running sessions instead of killing them silently (v2.1.203)
- [DOCS] Background agent working directory deleted, replaced by a file, or invalid: docs do not mention the v2.1.203 "fail once with a clear error" behavior
- [DOCS] Worktree docs never describe the multi-repo / nested-repository case that v2.1.203 fixed for `--worktree` and background-session isolation
- [DOCS] Worktree-isolated subagents docs do not state that shell commands now run inside the subagent's own worktree (v2.1.203)
- [DOCS] Bash tool docs omit the "argument list too long" limit and the v2.1.203 worktree-repo fix
- [DOCS] `claude agents` returning to the list does not document that in-flight background subagents now carry over
- [DOCS] `agent-view.md` "background service did not respond" troubleshooting never mentions the v2.1.203 automatic recovery when the daemon's session token goes stale
- [DOCS] `agent-view.md` version history omits v2.1.203 and no troubleshooting entry describes the macOS 15–20 second stall when opening or switching background sessions (regression fixed in 2.1.203, introduced in 2.1.196)
- [DOCS] MCP `roots/list` reference and `list_changed` section do not describe the v2.1.203 client-side update that exposes additional working directories and emits `notifications/roots/list_changed`
- [DOCS] Manual mode footer indicator: v2.1.203 grey ⏸ badge in the footer for manual permission mode is not documented
- [DOCS] [Authentication] No documentation for the v2.1.203 proactive login-expiry warning that prevents background-session interruption
- Feature: skill visibility tiers (deferred/lazy-loaded skills)
- [Feature Request] Add auto-update capability to Claude Code CLI
- Add option to strip/disable left gutter so copied multi-line output has no leading spaces
- [FEATURE] Mouse wheel scroll in fullscreen TUI only affects input/history region, not transcript pane
- Workflow agent() spawns inherit the session model with no override; built-in workflows ship unpinned — one deep-research launch ran 84 agents on Fable 5
- [Feature Request] Add option to disable click-to-answer while preserving text selection in chat
- MCP servers marked "still connecting" should transition to ready/failed/timed-out (session-startup determinism)
- Allow --resume from any directory (decouple session resume from origination directory)
- Keyboard-only scrolling not available in conversation view
- copier coller sur le texte de la réponse sur claude code???
- [Feature Request] Keep Claude 3.5 Opus 4.6 available as a model option alongside newer versions
- Let the Bash tool run commands under bash, not the user's login shell (zsh)
- Feature request: lightweight one-step commit+push command/UI
- [Feature Request] Restore direct skill invocation from agents view without initial message requirement
- Background daemon processes (bg-spare/bg-pty-host) show raw version string as process name
- dangerouslyDisableSandbox allows the model to bypass workspace boundaries without a distinct, explicit user confirmation
- [Feature Request] Add security and compliance code scanning capability
- [Feature Request] Add simulator UI automation support for iOS testing
- /insights: large-session facets record a 500-token error that never occurred, misfiling them as unclear_from_transcript
- [FEATURE] Machine-readable AskUserQuestion state and no-answer semantics
- [FEATURE] Sub-agent visibility
- Worktree cleanup can discard another session's in-progress work — no liveness signal, only git state
- [DOCS]
- [Bug] AskUserQuestion timeout instructs premature auto-proceed and is not configurable
- [BUG] No path-level exclusion for sensitive files in file-change-notification behavior
- Auto-mode's "Merge Without Review" rule blocks a signature-preserving `git push` to `main`, mischaracterizes a fast-forward as a force-push, and has no way to express repo-specific release policy
- [Feature Request] Sync terminal tab title with /rename command
- Expose extended (1h) prompt-cache TTL for interactive sessions
- [DOCS] Install script's new out-of-memory explanation needs a troubleshooting entry in the installation pages
- [DOCS] `axScreenReader` description lists only "decorative borders and animations"; v2.1.200 added glyph hiding, transcript labels, and nested-table-as-`Header: value.` rendering
- [DOCS] Fullscreen `fullscreen.md` tmux caveat still says synchronized output is unsupported; v2.1.200 enabled it for tmux 3.4+
- [DOCS] Voice dictation troubleshooting never lists the misleading "Voice connection failed" message fixed in v2.1.200
- [BUG] Auto-update daemon transition reaps a live session's in-flight headless background tasks (status 'killed', no attribution) — Linux, 2.1.210→2.1.211
- Feature request: bind keyboard shortcuts to custom slash commands in keybindings.json
- statusLine command: terminal_width missing from JSON input
- Fable 5 [cyber] safeguard false-positives make the model unusable for telecom/VoIP development
- [BUG] Desktop: "+ New" worktree session fails with "WorktreeCreate hook failed: path contains control characters" (repo not under a TCC-protected folder)
- [Bug] Tokens consumed but no output generated
- You've hit your session limit · resets 5:50am (Asia/Kolkata) · progress saved
- [BUG] claude-api skill inlines its entire ~800KB doc set when language detection fails, even on deliberate invocation
- [BUG] defer_loading + cache_control 400 still present in 2.1.223 (recurrence of #30920)
- [BUG] Bash tool timeout backgrounds a command instead of cancelling it — contributed to multi-hour VM disk-saturation incidents under concurrent subagent load
- [Bug] Extended thinking effort level not reflected in UI display
- [BUG] Session transcript repeats the full `usage` object for every content block of one API response (token over-counting; follow-up to #6805)
- Renaming a session from the /resume picker desyncs the input-border name (disappears or shows stale value after resume)
- [Bug] grep shell shim silently skips binary and ignored files, masking search failures from agents
- Claude/Chrome estension
- Embedded claude-code CLI re-triggers macOS 'access data from other apps' TCC prompt every new session
- Desktop app: embedded browser panel silently drops custom-scheme deep links (obsidian:// etc.)
- Fullscreen alt-screen TUI teardown freezes VS Code integrated terminal over Remote-SSH on exit
- [BUG] Windows: startup welcome banner only shows in fresh folders, not frequently-used ones (still repro on v2.1.193)
- API Error: 400 This organization has been disabled.
- Opus 4.8 (1M context) — frequent "tool_use malformed" client errors despite tools executing successfully
- Autocompact not triggering with Custom model (mimo-v2.5-pro) despite exceeding context limit
- Background agent notifications route through wrong agent ID when launched sequentially (2.1.172+)
- [BUG] Claude Code desktop app not showing Chat or Co-Work history
- [Bug] ESC key stops work instead of dismissing /btw message during execution
- [BUG] Unauthorised launch of extremely inefficient subagents that wasted tokens - I want a refund
- [Bug] Vim mode problems: 'n' keybinding for adding notes not functional
- [BUG] OAuth login/refresh fails with UNABLE_TO_GET_ISSUER_CERT on platform.claude.com's new Let's Encrypt (ISRG Root X2 cross-signed) chain
- Claude Code cannot be trusted: every response requires adversarial verification
- [Bug] Image pasting not working with clipboard (Ctrl+V, Ctrl+Shift+V)
- [BUG] What is happening?? Claude code went from hero to zero this week!.
- [Bug] Anthropic API Error: Content flagged as policy violation for backup operations
- [BUG]
- Subagent returns prompt-injection-shaped output with 0 tool uses (general-purpose, model: opus), localized to one long-running session
- Cowork not activating for PRO plan: cobalt_plinth feature flag disabled by policy
- [Bug] Auto-update fails with install_failed status
- [BUG] claude.ai marketplace MCP server instructions inject into context on every session regardless of Tool Search
- [BUG] VS Code sidebar omits sessions created via SDK/headless (entrypoint: sdk-cli)
- [Bug] Auto-mode spawns infinite monitor loops on batch classifier denial, causing runaway API usage
- [Bug] Concurrent user edits reverted during code reformatting
- [BUG] Commits authored by claude-code@anthropic.com are attributed to unrelated GitHub user
- [BUG] Background-task indicator (N shell chip) invisible when theme background token matches terminal background
- [Critical][Safety] Idle /rc bridge leaves a session live and forks it into an uninterruptible autonomous process - writes files & controls hardware with no operator kill path (v2.1.193)
- [Bug] TUI scrollback buffer insufficient for long responses
- [Bug] Math calculation error in pricing model analysis
- VSCode extension: assistant tool calls rendered as literal text in chat, not executed
- [BUG] VSCode extension ignores claude environment settings
- [BUG] Got a session limit issue without actually using it, hit the limit exactly at the start of the reset
- [BUG] VS Code Chat panel: "Error during execution" when resuming a previous chat session (Windows, v2.1.196)
- Background task chips persist as 'Running' after subagent exits; Stop button no-ops
- Voice mode fails in terminal CLI: WebSocket 1006 close and unstructured mic error
- Background Bash runner process (re-exec'd versioned binary) leaks memory after command completes — 100% CPU, OOM-killed at 10.8GB (v2.1.226 & v2.1.233, Linux)
- Per-tool-call helper process (claude.exe) leaks to 11.6GB anon RSS and is OOM-killed during a slow sandboxed Bash command
- Skill tool substitutes invocation args into literal "$0" text inside SKILL.md
- Delivery succeeds, execution does not
- [BUG] the progress sectio is not completely closing.
- /claude-api skill loads the entire multi-language bundle (~230k tokens) instead of the detected project language
- [BUG] claude code skill unconditionally inlines all files under shared/
- [BUG] /claude-api bundled skill saturates context unconditionally — ~77% spike from neutral question
- [BUG] Chrome file_upload rejects every path in scheduled-task sessions (worked before, breaks weekly business workflow)
- [BUG] Embedded ugrep hangs indefinitely (pegged core) on unanchored bounded-repetition pattern — the same binary's embedded ripgrep finishes in 53 ms
- Permission rules: $HOME not expanded during startup validation when checking symlinked config
- [Bug] Remote control functionality not working
- Claude in Chrome not working
- [Feature Request] Add safeguard disable option or allowlist for high-volume API users