Digest / September 17, 2026
Claude Code Issues Digest — September 17, 2026
358 new issues filed ·
364 resolved in anthropics/claude-code (UTC day).
New issues (358)
- [FEATURE] Desktop app: add /branch and /resume equivalents for managing conversation branches
- Claude in Chrome extension never connects (tabs_context_mcp always returns "not connected") despite reinstall + full browser restart
- DiviOps MCP server subprocess fails all outbound fetch calls in the desktop app's Code tab specifically ("Connection failed: fetch failed") — works fine from the standalone terminal CLI
- [Bug] MFA verification fails with "Could not verify the account security policy" error
- [FEATURE] Desktop Code sidebar: show project folder next to session title when Group by: None
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- [BUG] Scheduled session git clone fails with 401 — proxy git credential auto-injection (gitConfigInjection) not working
- [FEATURE] No setting to disable unused beta tool schemas (Workflow/Artifact/DesignSync/Cron/worktrees) inflating context
- [Feature Request] Add setting to disable automatic selection attachment in terminal mode
- [BUG] Home/End in multi-line prompt input jump to buffer start/end, not line start/end; Chat context has no line-navigation actions to rebind
- [BUG]
- Workflow tool + Ultracode guidance produced ~137 agent calls and zero usable output on a low-stakes task
- [Bug] Anthropic API Error: Incorrect cyber classifier flagging legitimate streaming/multimedia parameters
- [Bug] Model ignores output style directive, generates voseo Spanish despite neutral tuteo configuration
- [Bug] Docker Compose project-name collision causes `down` to remove unrelated containers
- [FEATURE] Add "Copy" button under assistant responses (like Zed's AI panel)
- Mobile app hides remote-control sessions until Claude GitHub App is installed; web banner stale after install
- [Bug] Ctrl-based keyboard shortcuts unresponsive in agent view after extended use
- [FEATURE] Add API/webhook (event) trigger for Claude Cowork tasks — not just schedule cadence
- Bug / Authentication Error.
- Feature request: Tabbed/multi-project sessions with a project sidebar
- Withdrawn by the author
- Withdrawn by the author
- I appreciate the context, but this doesn't appear to be a technical bug report for Claude Code. This reads as personal feedback about an agent's behavior and decision-making patterns rather than a software issue. If you'd like to report an actual technica
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- [FEATURE] Let /clear reset context without wiping the terminal scrollback
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- VS Code: Clicking sidebar session doesn't switch to existing tab, overwrites current view
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- [BUG] @ file mentions cannot reference files inside nested git repositories (desktop Code tab)
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- [BUG] Unrecognized slash commands moved from a free, deterministic rejection to a paid, highly variable model call, with cost rising undocumented across releases
- [BUG] Remote Control session stuck disconnected after auth token missing required scope, no way to reconnect from UI
- Subagent `tools:` entry with a Bash specifier grants unrestricted Bash instead of scoping it
- I need a bug report to generate a GitHub issue title. Could you please provide the bug report details?
- Claude in Chrome: CLI never contacts the native messaging host (Status stays "Disabled")
- [Bug] Prompt submitted during active request misinterpreted as tool result block
- [FEATURE] Make the Claude iOS app (Claude Code mobile client) available on Apple Vision Pro
- [Bug] Claude stops execution without cause and ignores loop conditions
- [BUG] Claude Code is showing one model selected while running another
- Windows/Git Bash: session's Bash tool dies permanently (unalias: unsetenv, then "completed with no output") and does not recover on resume
- [Bug] Image clipboard detection fails despite valid image present
- Channels (Telegram): Claude doesn't automatically call the reply tool for inbound channel messages — responds only in terminal
- Fable 5.1 safeguard false-positive [cyber] — coordinator session blocked while reading routine status reports about hardening our own app's permission checks (req_011Cf8PR5egA5gvAG6WF3LLW)
- [Desktop, Windows] File links outside the session folder get no file menu, and no pane can show them
- [Bug] Cyber guardrails incorrectly flag approved cybersecurity tasks after verification
- [Bug] macOS Finder image paste attaches file icon instead of actual image
- claude mcp add reports success when the config write is denied (sandboxed Bash)
- [Bug] Codex Reviewer runs at low reasoning effort for adversarial code review
- [BUG] MCP session-expiry recovery retries `tools/call` only; `readMcpResource` still fails once with `Connection closed` after a Streamable HTTP server restart
- [BUG] FORKING
- [BUG] `claude mcp list` ignores allowedMcpServers/deniedMcpServers — empty allowlist still returns every server connected
- [BUG] /goal evaluator schema rejected by strict backends: required omits impossible
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- /clear and "Clear conversation" create a new empty session instead of clearing the current one
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- [FEATURE] settings.json key to fully disable IDE selection/open-file context injection
- Withdrawn by the author
- Withdrawn by the author
- [BUG] Claude Desktop 2.110.0 (Windows/MSIX): after a quit, every launch fails with renderer "launch-failed, exitCode: 21" until CoworkVMService is restarted
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- [BUG] I am unable to see my research history on the desk top app in mac. but is all available on web.
- [FEATURE] Desktop app: show a local/remote (SSH host) indicator on projects and sessions in the sidebar
- [Bug] Opus 5 reasoning_extraction safeguard blocks benign first message, recurs across fresh sessions (Windows)
- [BUG] When drag out the chat tab it looses the prompt
- Desktop: "Couldn't open that session" error when running /desktop before sending a message
- [FEATURE] Feature request: easy way to move a session from Cloud to Local (Local→Cloud nice-to-have)
- Orphaned advisor_tool_result after /compact makes the session permanently un-continuable (400)
- [Bug] False positive security detection triggered without malicious content in file
- Workflow resumeFromRunId re-runs every completed agent after the first parallel fan-out (40 redundant opus agents, 231M tokens)
- [Bug] Anthropic API Error: Fable 5.1 safety filter triggered on HTML escaping operations
- [BUG] claude spins at 100% CPU indefinitely on startup — regression between 2.1.112 and 2.1.113
- Auto-memory: model repeatedly writes memory files for inconclusive/low-value tasks
- [Bug] Anthropic API Error: Model claude-opus-5.0 not available for Cyber Verification Program
- Ctrl+C intended to clear the prompt input killed 6 running background agents (no confirmation)
- [FEATURE] Warn before Console API credits expire, the way subscription usage credits already do
- Path-scoped .claude/rules only load via the Read tool, while auto mode instructs the model to prefer Bash
- [Bug] Plugin updates fail silently and version visibility is poor
- iOS Simulator control tool: attach/boot times out on Xcode 27 — targets legacy Simulator.app, which no longer exists (replaced by DeviceHub.app / com.apple.dt.Devices)
- Desktop app: send_message to a busy session waits for the turn to end, while typed messages are picked up mid-turn
- [BUG] tool_addition blocks are sent to a custom ANTHROPIC_BASE_URL, rejected with 400, and silently retried on every session start
- [BUG] Chrome cookie import fails: Safe Storage key queried in data-protection keychain, but Chrome stores it in the legacy login keychain (macOS)
- [Feature Request] Consider confirmation prompt before making unrequested additions
- [Feature Request] Pin latest prompt to terminal top for improved multitasking visibility
- Feature request: tidy, named tab groups for Claude in Chrome
- Desktop app: diff view lacks persistent defaults (uncommitted-only mode, collapsed files)
- I can't use my claude code
- [Bug] Cyber Verification Program members experiencing workflow blocking
- [FEATURE] Plugin-shipped CLAUDE.md: stable path or @import expansion — the stale-pin failure #82068 predicted, now observed
- [Bug Report: Performance regression and increased token usage in Claude Code
- [Bug] /context command performs slowly with >1m latency
- [BUG] Desktop app chat sidebar stuck empty after forced re-authentication (right after chat/collaboration merge announcement) — data confirmed intact on claude.ai
- [Bug] Session persistence lost after account renewal
- [BUG] Marketplace name is derived from the bare repo name, so any two `*/skills` repos collide
- [Security Research - Authorized] Automation reproduction case
- ralph-wiggum stop hook: jq error handler is unreachable under set -e, leaving a stale ralph-loop state file
- False "Update available! Run: mise upgrade claude" when mise is already on latest
- [BUG] nested_memory injects a directory's .claude/rules/* but skips that directory's CLAUDE.md
- Remote (cloud) session: project skills from .claude/skills are missing from the / slash-command menu
- [DOCS] Clarify whether Recently denied approval is one-time or persistent
- [BUG] Slash-command menu lists a plugin command the session cannot run: the menu reads the marketplace source, the runner reads the cache copy
- [Bug] Anthropic API Error: Model access denied for claude-opus-5
- [Bug] Claude 3.5 Opus 1M model stops responding after initial task completion
- [Bug] Excessive API token consumption from instruction non-compliance
- [BUG] Claude Desktop (Windows MSIX): main webview renderer fails to spawn — render process gone: launch-failed, exitCode 21
- [BUG] MSIX build: claude.exe / Cowork VM bundle / VM-internal claude binary all fail to land after download — ENOENT on final rename
- MCP OAuth credentials persisted with expiresAt: 0 and no refreshToken, forcing re-auth every session (mcp.grafana.com, mcp.slack.com)
- [BUG] VS Code extension (Linux/Wayland): Alt+Tab back to a focused Claude Code panel leaves the VS Code menu bar focused, as if Alt had been pressed alone
- [BUG]
- [FEATURE] PermissionRequest hook input should include tool_use_id (per-call id) to correlate identical simultaneous prompts
- [BUG] Auto Mode safety classifier times out with claude-opus-5[1m], blocking Bash
- [Bug] Incorrect terminology: "spelling" used instead of "case" for case-sensitive string differences
- Abre pantalla completa y no me deja ni siquiera ver la hora
- [Feature Request] Make folder open behavior configurable to support custom applications
- Worktree isolation refuses git-free Bash on 2.1.272/2.1.273: literal-assigned variables, $(…) args, loops, heredocs after && (75 refusals in 3 days)
- Remote Control From Mobile Phone
- [BUG] Desktop app re-runs `git ls-files --others :/` ~2.2×/sec without single-flighting, burning ~3 cores to find 3 untracked files (macOS, large monorepo)
- Desktop app: option to make Enter insert a newline and submit only via click / Ctrl+Enter
- [Feature Request] Clarify browser code entry during terminal login flow
- [BUG] WebFetch truncation is invisible to the model — absent from the tool description, unmarked in the result, and the web-fetch subagent can neither detect nor recover from it
- [BUG][bedrock] /model picker's Fable 5.1 quick-pick writes wrong Bedrock ID -> 400 invalid model identifier
- [BUG] Claude Code CLI (`claude -p`) burns ~15-16k tokens of fixed overhead per call when used as a stateless batch text-completion backend
- Scroll broken (mouse wheel → arrow-key input) for sessions opened via Agents view; persists across terminals on reattach
- [Bug] Classifier incorrectly blocking priority planning requests as biometric data
- Regression: new session always opens in a separate/minimized window instead of the active editor's main tab area
- [Bug] Daemon spawns workers with a hardcoded 120x30 PTY when the client size is not yet known, then never resizes them (Windows)
- [FEATURE] Expose `model` and reasoning `effort` as settable properties on create_scheduled_task / update_scheduled_task
- sandbox credentials.files mask produces no file at all for ~/.config/gh/hosts.yml
- [Bug] Code-review subagent posts GitHub PR review without user confirmation
- [Bug] Significant performance degradation in Claude model responses
- Stopping a background agent does not stop the agents it spawned, and neither the stop result nor the task-notification mentions them
- [BUG] Browser pane still blocks same-origin subresources on DDEV's *.ddev.site (public DNS → 127.0.0.1) after #86362 fix; *.test works
- [BUG] Claude can no longer control MacOS Simulator after 27 update
- [BUG] Claude does delete conversation data after cleanupPeriodDays
- [Bug] Claude Code ignores claude.md instructions and delivers low-quality untested code
- [Bug] Anthropic API Error: Message Blocked by Claude Safeguards for Sports Statistics Processing
- [FEATURE] Desktop: keyboard shortcut for "Reply" on selected transcript text (e.g. Cmd+E / Cmd+Shift+R)
- [Bug] Model context pollution allows circumvention of refusal policies across model switches
- [BUG] Claude Desktop hangs and shows white screen on startup when mapped network drives (VPN) are unreachable
- Scheduled tasks with "이 컴퓨터 필요" toggle fails on all folders (macOS Sequoia)
- Session not appearing in sidebar despite valid session file on disk
- Render MCP Apps (SEP-1865) ui:// resources in Claude Code
- [Bug] Prompt injection in tool output misattributed as user instruction bypasses approved plan
- [BUG] /diff turn view: an image-only prompt does not start a turn, so its edits are folded into the previous turn or dropped
- [Bug] Anthropic API Error: Content filtered by safety policy (reasoning_extraction)
- Autostart setting is re-enabled on every Claude Desktop update, even when previously disabled
- [BUG] Fork subagent performs write actions despite parent session being in plan mode
- [BUG] Fork subagent performs write actions despite parent session being in plan mode
- [Bug] Dangerous git checkout command wipes user files without confirmation
- [BUG] Account-synced plugins cannot be removed: they reappear after removal from the desktop app AND claude.ai, and are invisible to the CLI
- [BUG] Claude in Chrome: phantom/disconnected tab state persists across full Chrome + app restart (macOS desktop app)
- [BUG] Scheduled task hangs indefinitely on its first Bash tool call when run unattended (no timeout, no error)
- I appreciate the message, but this appears to be a request for increased API limits or account privileges rather than a bug report for Claude Code. If you're experiencing actual issues with Claude Code (the CLI tool), please provide: - What specific error
- [FEATURE] "Update branch" action in the desktop app's CI monitoring popover
- [Bug] Anthropic API Error: False positive cyber safeguard block on tool completion messages
- Feature request: show git worktree in the agent list (claude agents / FleetView)
- [FEATURE] Group sessions in the agents view by tag (/tag, tag column, collapsible groups)
- /code-review in the desktop app runs inline in the authoring session (no context isolation), and the docs don't say so
- [FEATURE] Per-server "always allow" for locally installed MCP servers (persisted across sessions and chats, revocable)
- [Feature Request] Add command to view last message in chat history
- [BUG] [VS Code / Cursor extension] Manual session tab rename failure makes >= 2.1.269 not usable
- Copy icon on a response should include images, not just text
- [Bug] Unauthorized file system access beyond explicitly permitted directories
- [Bug] Usage policy violation errors persist after /clear command
- [Bug] Session crashes or becomes unusable
- [BUG] Remote Control failed to connect: Remote Control server rejected the request (HTTP 403)
- [Bug] Arrow key navigation broken in "/btw" menu history
- Plugin marketplace auto-update and manual refresh both non-functional (autoUpdate: true never fires)
- [Bug] Unauthorized model switching during authentication server development
- [BUG] Claude Code sends the argument text of a stacked-skill prompt two times per skill (six copies of a pasted document for three skills)
- Google Drive connector: no granular editing for Docs and Sheets
- gitlab mcp problems reading joblog
- [BUG] Desktop (SSH): "Worktree location" accepts only a local path, so remote sessions always fall back to the remote $HOME
- [Bug] Auto mode blocks local SendMessage without clear classification reason
- [Bug] Ultra Mode Functionality Regression
- [Bug] Zap scan blocks incorrectly during execution
- MCP server listed as "Connected" in claude mcp list (CLI) but missing from session's connector list in Claude Desktop
- [BUG] Predicted-prompt autofill synthesizes destructive commands against targets the user never mentioned
- code-review skill: nested-fork not supported at high+ effort causes runaway sub-agent duplication
- VS Code extension: tab silently attached to a different session, no user action
- [BUG]
- [Bug] Incorrect Safety Guard Trigger
- [FEATURE] Choose model and effort per task when suggesting a chip (spawn_task) or starting a session
- [Bug] iOS Simulator input actions fail with sandbox: "Failed to initialize IO ports for device IO client"
- VSCode extension "Web" tab doesn't show cloud session created from mobile app
- [Bug] Error handling for long string values in input parameters
- [Feature Request] Auto-compact should trigger immediately when enabled mid-session with full context window
- [FEATURE] Desktop sidebar: show sessions as a parent/child tree
- [FEATURE] Interactive /statusline picker: choose, reorder and live-preview built-in status line components
- [Bug] Agent fails to resume task execution after standby mode
- Add a PreClear hook event that fires before /clear wipes context
- Auto mode regression since 2.1.270: classifier blocks a solo owner's own routine release work (12× more denials), and the Manual fallback costs 55+ permission clicks for a 2-key config change
- [BUG] Assistant repeatedly acts without permission despite explicit, repeated, written instructions not to — full session audit with exact quotes
- Subagents keep polling long jobs with shell sleep loops instead of using run_in_background/Monitor notifications
- [BUG] Session history silently deleted with no warning when app is reopened after 30+ days
- [FEATURE] Título: Voice dictation: auto-send message by saying a trigger word (e.g. "enviar" / "send")
- [BUG] Claude Code from android app showing always question popup even after answered
- [BUG] opusplan silently stops upgrading to Opus in plan mode once conversation context exceeds 200k tokens
- Bash-tagged Run-button code blocks with cross-drive cd on Windows fail silently depending on which shell executes them
- [BUG] PowerShell(git commit *) allow rule never suppresses the confirmation prompt, while other git subcommands work
- [Bug] Safeguards triggering unexpectedly on legitimate anti-fraud system code
- I appreciate you sharing this context, but I need to clarify something important: **this appears to be a message intended for Anthropic support or a policy discussion, not a bug report for Claude Code.** This isn't related to Claude Code (the CLI tool), i
- Stop button semantics are unpredictable: sometimes work keeps going, sometimes it silently kills all background agents
- MCP tool-call validation rejects omission of any optional/defaulted parameter, regardless of type
- Model confirmed then defended a wrong design rationale across turns; perceived reasoning regression on claude-fable-5-1
- [BUG] claude.exe attempted to access Firefox cookie database (blocked by Norton)
- [BUG] /feedback: send does nothing, review is truncated, and "0 to dismiss" turns the feature off
- [Bug] ExitPlanMode opens blank approval dialog
- Headless service account cannot run `claude -p`: silent exit 0 when the OS user has no onboarding state, despite `auth status` reporting `loggedIn: true`
- [FEATURE] Prioritize /status over /stop in autocomplete for /st
- [Feature Request] Implement lazy-loading of connector tool schemas and Tool Access Mode settings
- Assistant fabricates "user" turns — seems tied to unnecessary AskUserQuestion tool-search calls
- [Bug] Claude Code fails to generate coherent 2 sentence responses
- [BUG] Expired Prompt Cache Causes Large Token Re-Caching and Disproportionately Consumes 5-Hour & Weekly Usage Limits in VS Code Agent Mode
- Add a quick thumbs up/down feedback control to Claude Code sessions
- [Bug] Claude Code ignores user prompts and generates irrelevant code implementations
- [Bug] High Token Consumption with Incomplete Task Execution
- Tool-approval annotation text is not applied to the in-flight tool call
- VS Code webview does not render the Task* tool family (progress/todo panel stays empty)
- [BUG] VS Code extension: AskUserQuestion option `preview` is silently dropped by the webview — options render as label + description only
- [Bug] Chat history not persisted correctly, lost 40+ messages after training
- [BUG] share artifact android
- [BUG] Remote Control sessions are hidden from the local /resume picker (filtered as entrypoint=sdk-cli)
- [Bug] Opus 5 with medium effort mode shows exponential latency and token rate increase
- [BUG] managed-settings.json sandbox.network.deniedDomains not enforced for Bash tool in VS Code extension
- Desktop app: messages written after forking a session are hidden in the viewer and omitted from Export (per-row sessionId mismatch)
- "Save skill" button does nothing when clicking a packaged .skill file card
- [BUG] Windows: `.oauth_refresh.lock` directory survives a mid-refresh exit and is never reclaimed despite `stale: 60000` — every `claude -p` call fails until it is deleted by hand
- Claude Code Desktop: Support multiple persistent inline annotations on a Claude response before sending
- Auto mode's Bash-first instruction is unusable with permissions.blockReadsOutsideWorkingDirectories
- MCP tool-call validation treats optional/nullable parameters as required, and literal "null" is never transmitted as JSON null
- [BUG] CLAUDE_CODE_EFFORT_LEVEL in .claude/settings.local.json latches at session launch; later writes, including `unset`, are read and discarded
- "+ New" doesn't prompt for a project folder — easy to start a session in the wrong project
- Monitor/TaskStop removes the task from the panel but leaves the spawned process running (orphaned tail -F accumulates)
- [Bug] Overly aggressive content filtering on git commit messages and SQL queries
- Worktree pool reuses worktrees that belong to live, unarchived sessions — and evicts the owning session into the origin repo
- Desktop app: idle worktree reaper silently deletes a session's working folder; session then shows 'Working folder no longer exists' with no explanation
- [Bug] Windows path length limit causes file deletion failures and token loss
- false positive message
- [Feature Request] Support for local LLM models as alternative to Anthropic API
- [BUG] theme: "auto": message blocks keep the colors of the theme they were drawn under; a live appearance switch never repaints them
- [Feature Request] Allow running jobs to complete when credit limit reached instead of immediate termination
- Allow remapping/disabling GUI menu keyboard shortcuts (View menu, etc.)
- VSCode extension: Session History panel empty/stale due to missing or orphaned sessions-index.json
- Host-side AffixIO attestation for privileged tool calls (PreToolUse / PermissionRequest)
Resolved issues (364)
- [Bug] MFA verification fails with "Could not verify the account security policy" error
- [Bug] Anthropic API Error: Incorrect cyber classifier flagging legitimate streaming/multimedia parameters
- [Bug] Codex Reviewer runs at low reasoning effort for adversarial code review
- [Bug] Generated examples lack logical coherence with prior context
- [Bug] Last prompt not persisting after v2.1.251 update
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Withdrawn by the author
- Custom MCP connector (SSE, OAuth+DCR+PKCE) authorization fails on Claude Web/Desktop after server issues valid token — CLI succeeds with identical flow
- [Bug] Anthropic API Error: Model claude-opus-5.0 not available for Cyber Verification Program
- [Bug] Session persistence lost after account renewal
- [Security Research - Authorized] Automation reproduction case
- [Bug] Anthropic API Error: Model access denied for claude-opus-5
- 상단 툴바의 코드(</>) 버튼에 텍스트 라벨이 사라지고 클릭해도 반응 없음
- [Bug] Repeated API errors require multiple correction attempts
- [Bug] Goal loop not always triggered when setting goal from desktop/mobile
- [Bug] Claude Code defaults to Opus 4.8 instead of respecting specified model preference Fable 5.1
- [Bug] Unexpected High Usage Consumption and Rate Limiting Issues
- [Bug] Unexpected tool execution on user query without explicit request
- [Bug] Invalid Input: Unable to Process Request
- [Bug] False positive detection causing premature exit from agentic loop
- [Bug] Agent generates responses from memory instead of verifying code artifacts for critical design issues
- [Bug] Vague or unclear API responses from Claude model
- [Bug] Flagging system usability issues
- [FEATURE] defer_loading for Managed Agents
- [Feature Request] Add support for machine learning experiment workflows in Claude Code
- [Feature Request] Add support for authorized security testing workflows
- Voice/TTS playback for Claude Code sessions in the mobile app
- [FEATURE] Show background agent output in separate tab
- Routines: allow naming a run when firing the API trigger
- [FEATURE] Recognise Microsoft Global Secure Access synthetic IPs (`6.6.0.0/16`) for gateway `/login`
- [FEATURE] Add in-memory storage option for background task outputs to prevent sensitive data leakage to disk
- [FEATURE] Let recipients propose edits to a shared skill, with owner approval and automatic republish
- [FEATURE] Managed / Remote settings have poor layering and are trivially bypassable
- Support Claude Desktop-style interactive OIDC for MDM-managed third-party gateways in Claude Code
- Feature request: automatic context clearing (in addition to auto-compact)
- [Feature Request] Improve voice model training capabilities and documentation
- [Feature Request] Add security audit/hardening assistance capabilities
- [BUG] Settings schema: `strictKnownMarketplaces` rejects `source: "settings"`, which the runtime accepts and enforces
- [Feature Request] Add token pricing adjustment for politeness in prompts
- [Feature Request] Add agent intrusion detection and logging capabilities
- [FEATURE] Parameterized plugin-root placeholder `${CLAUDE_PLUGIN_ROOT:<plugin-name>}` in settings.json hooks
- [FEATURE] Disable automatic File Reference in Claude Code
- [Feature Request] Privacy and Security tab not visible in UI
- [FEATURE] Feature request: allow deleting/editing a specific portion of the conversation history Cuerpo:
- [Feature Request] Improve documentation clarity for formatting rules and styling conventions
- [Feature Request] Per-source token usage visibility and anomaly detection alerts
- [FEATURE] : Organize routines into user-defined groups
- [Feature Request] Add diff view for plan changes and separate plan sections by mode switch
- [FEATURE] Claude code agent selector feature request
- [Feature Request] Improve safety classifier context sensitivity for authorized GRC/ISMS documentation tasks
- [Feature Request] Make /code-review command agent-invocable
- [FEATURE] Desktop composer reserves a full-height right gutter, costing ~4 characters per wrapped line
- reserve % of quota for specific tasks
- [VSCode Extension] Support drag-and-drop of images/files into chat input
- [Feature Request] Add read-only access to Claude Desktop conversation history for Claude Code integration
- [Feature Request] Relax constraints for general code review workflows
- [FEATURE] Let cloud sessions comment on public repos the user cannot push to
- [Feature Request] Reduce verbose metaphorical language in Claude Code responses
- Claude stopped before I told it to, and it was right — please instrument that decision (context budget per task)
- Tiles in the Background Tasks tab should show start/finish timestamps
- [FEATURE] Start new Remote Control sessions in any git subdirectory of a published root, instead of one server per pre-registered directory
- [Feature Request] Document statusline usable width and fix right-aligned element truncation
- [FEATURE] design-sync for react native
- [Feature Request] Reorder Shift-Tab permission mode cycling to manual-accept-auto-plan sequence
- [Feature Request] Add option to pre-populate editor buffer with previous assistant message as context
- Feature request: plan-approval parity on remote-control surfaces — offer "accept, clear context and auto mode"
- Feature request: separate portable configuration from machine-local state under ~/.claude (or ship a recommended .gitignore)
- [FEATURE] Stage questions with their captured context while Claude Code is working, review the queue before it's consumed
- [FEATURE] Allow subagents to inherit the parent's prompt cache for shared dynamic context
- [FEATURE] Allow relocating a Cowork project's primary folder without losing project data
- Feature request: configurable height/row count for the inline agents tray
- I don't have enough information to generate a meaningful GitHub issue title from your message. Could you please provide: 1. A clear description of the bug or feature request 2. What you were trying to do 3. What error occurred (if any) 4. Expected vs actu
- [Feature Request] Model-directed selective context eviction (mark-and-sweep), as an alternative to whole-history compaction
- [Feature Request] Support for lab/sandbox environment configuration
- [BUG] macOS DarkWake: transient DNS failure during OAuth refresh loses login
- Feature request: inline reply/annotation with numbered references for chat responses
- [FEATURE] Remote Control: same-name re-registration should supersede the old session entry (disconnected duplicates accumulate indefinitely)
- [FEATURE] "Uploading" visual clue
- [FEATURE] A port-allocation coordination primitive for concurrent sessions (re-file of #34385)
- [FEATURE] Screen sharing feature
- Feature request: allow AskUserQuestion to render inline in the response text, not just as a modal
- [Feature Request] Support for Fable 5 model
- [FEATURE] Deep link to resume an existing Cowork session (claude://cowork/{session-id})
- [FEATURE] Define precedence when CLAUDE.local.md conflicts with project CLAUDE.md instead of arbitrary selection
- [FEATURE] Add a progress bar for `claude update` and `claude upgrade` commands.
- [VS Code extension] Option to collapse/hide tool-use cards and their output in the chat feed
- [FEATURE] Desktop local sessions: user-facing scoped sandbox and approval controls
- [Feature Request] Add support for Notion-to-GitHub migration workflow
- [FEATURE] Support .claude/skills.local/ for project-scoped, uncommitted skills
- Quick model and effort change by numbers
- [FEATURE] Auto plan mode
- [FEATURE] Workflow tool: no fail-fast on rate-limit/quota exhaustion across concurrent subagents
- Feature request: stronger visual distinction between user prompts and assistant output
- [FEATURE] Compress session transcripts on disk (~65% reclaimed)
- [FEATURE] Desktop: replace the delete-session confirmation modal with a recoverable "Recently deleted" view
- [FEATURE] Expose typed per-agent cancellation handles in dynamic workflow JavaScript
- [Feature Request] Add "Jump to Top" navigation for latest response in mobile app
- [FEATURE] Subagent model policy: relative bias + cap/floor bounds applied after the resolution cascade
- [FEATURE] Make images pasted into prompts clickable
- Feature request: read-only usage scope on `claude setup-token` (or a usage:read grant) for subscription accounts
- [FEATURE] Per-call timeout override for AskUserQuestion (not just the global askUserQuestionTimeout setting)
- [FEATURE] Declarative plugin config: non-interactive install --config + plugin config get/set (config-as-code)
- [Feature Request] Custom verb conjugation for completion messages
- Feature Request] Support for unavailable models in Claude Code
- [FEATURE] Expose agentType per task in the subagentStatusLine payload
- [FEATURE]
- VSCode extension chat input doesn't honor ~/.claude/keybindings.json — causes accidental submits
- Support model selection via text prompt syntax (e.g., /model sonnet-5)
- [Feature Request] Add Model-Invocable RenameSession Tool for CLAUDE.md Auto-Titling
- [FEATURE] Support plan mode with Claude Code Channels
- Desktop Feature Request: In-window preview pane for native macOS apps
- [FEATURE] Usage refresh button in claude mobile app
- [Feature Request] Improve safeguard handling for legitimate chemical/toxicology research contexts in LCA software
- Feature request: Dispatch — choose folder (preset working directories) and permission mode when spawning a Code session from mobile
- [FEATURE] Auto-expand enum option lists in MCP elicitation form dialogs
- Plan mode needs a discuss/debate state that doesn't wipe the plan view
- Agent should stop and report, not work around, a failed Docker context connection
- Prohibited block forbids all deletion, unscoped — blocks rm, log scrubbing, and cleanup in coding/ops agents
- [BUG][bedrock] /model picker's Fable 5.1 quick-pick writes wrong Bedrock ID -> 400 invalid model identifier
- [FEATURE] Expose `model` and reasoning `effort` as settable properties on create_scheduled_task / update_scheduled_task
- [BUG] Fork subagent performs write actions despite parent session being in plan mode
- [Bug] Session crashes or becomes unusable
- Subagents keep polling long jobs with shell sleep loops instead of using run_in_background/Monitor notifications
- [BUG] Cowork task starts with no device_* tools while the bridge is authenticated (Windows 10 22H2, after KB5129236)
- Model confirmed then defended a wrong design rationale across turns; perceived reasoning regression on claude-fable-5-1
- [BUG]
- [BUG] Claude Pro payments fail before issuer authorization; support human escalation loops for 1+ month
- I need more information to generate an appropriate GitHub issue title. "Stay ahead" doesn't contain enough detail about a specific bug or feature request. Please provide: - What problem or bug are you experiencing? - What error message (if any) are you se
- Claude Desktop "Open" attach picks wrong file content despite showing correct filename in confirmation dialog
- Knowledge cutoff leaves everyday answers stuck in early 2026; must pay for a web search on every current question
- [Bug] Session safeguards preventing code modifications
- [Bug] Source code obfuscation/protection features blocked
- [Bug] Anthropic API Error: Safeguard False Positive on Legitimate Code Task
- [Bug] Agent fails to follow instructions and performs shallow analysis
- [Bug] Claude Opus 5 safety filters triggered during proxy hardening configuration
- [Bug] Agent spawning excessive sub-agents for simple tasks
- I appreciate you reaching out, but I'm not able to help with that request. If you have a legitimate bug report or feature request related to Claude Code (Anthropic's CLI tool), I'd be happy to help you create a GitHub issue title for it. If you're looki
- [Bug] CWR false positive flagging non-weapon mod as weapon
- I don't see a bug report in your message - you've only provided a request ID. Could you please share the actual bug report or error details that you'd like me to create a GitHub issue title for?
- I don't have access to the specific bug report details for request ID `req_011Ceuq83GJvp79eAHeBxCkX`. To generate an appropriate GitHub issue title, please provide: 1. A description of the bug or feature request 2. Any error messages encountered 3. The
- I don't have a bug report or technical issue description in your message to generate a GitHub issue title from. Could you please provide the actual bug report or technical problem details for Claude Code? I'll need information such as: - What error occu
- [Feature Request] Allow security hardening and vulnerability assessment workflows with Claude models
- [FEATURE] Add Bedrock Guardrail to Claude Gateway
- [FEATURE] Bridge local stdio MCP servers into cloud Cowork sessions
- [FEATURE] Browser pane: no way to bypass certificate errors for local development hosts (self-signed certs)
- [FEATURE] Nextcloud Contacts connector for Cowork
- [FEATURE] Nextcloud Calendar connector for Cowork
- [FEATURE] Thunderbird connector for Cowork
- [FEATURE] Generic CalDAV calendar connector for Cowork
- [FEATURE] Generic IMAP email connector for Cowork
- [FEATURE] Nextcloud Deck connector for Cowork
- [Feature Request] Enable two-way advisor with model fallback for simple requests
- [Feature Request] Add visual distinction for queued messages in CLI output
- [Feature Request] Expose workflow concurrency override for high-resource hosts
- Show absolute timestamp (like WhatsApp) instead of relative "X minutes ago"
- Desktop: opening a session anchors mid-conversation instead of first-unread / bottom
- [FEATURE] Structured session journal: a configurable event stream with repo and OTLP sinks, detail levels, and privacy profiles
- [FEATURE] Add auto-send after silence timeout for desktop app voice dictation
- [FEATURE] Conversation anchors: re-enter a key message repeatedly, with parallel (shared code) and alternative (isolated code) branch modes
- [FEATURE] Re-attach project-scoped .mcp.json MCP servers when a conversation is resumed
- [Feature Request] Expose remaining budget and a configurable override for the 1000-agent-per-workflow lifetime cap
- [FEATURE] /simplify should cut AI slop comments
- [Feature Request] Enable voice mode support in remote control sessions
- [FEATURE] Support shallow/partial/sparse clone for large repositories in cloud environments
- [FEATURE] TUI: numbered transcript blocks + per-block action menu (Reply / Repeat / Restore-here) — reviving the #64207 reply-to-block lineage
- [Feature Request] Support for security research and academic experimentation modes
- [Feature Request] Add Fable model plan option for token-efficient task execution
- Group / nest related sessions in the Recents list (tree or "project" view)
- [FEATURE] Add an opt-in reality-filter / anti-sycophancy protocol with multi-agent authority boundaries
- [FEATURE] Side channel for CLI tools to emit context information
- [Feature Request] Reduce false positive safety flags for home automation use cases
- [Feature Request] Add process monitoring and analysis capabilities for system diagnostics
- [FEATURE] Reveal file in explorer when showing edit diff
- Feature request: de-emphasized/subtle text support for assistant-generated markdown
- [Feature Request] Add workflow file system report for caching support
- [Feature Request] Overly restrictive security flagging for routine cryptographic upgrades
- [FEATURE] Disambiguate project names in sidebar: show owner/repo or allow custom display names
- [FEATURE] voice-driven agentic control
- [FEATURE] Optional local RDP exposure check + brute-force attempt summary (Windows, opt-in)
- [FEATURE] VS Code extension: /compact summary output is not copy/paste-able
- Feature request: alphabetical or configurable sorting for the slash-command autocomplete dropdown
- [Feature Request] Add Roman Urdu language support to Voice mode
- [FEATURE]Version-pinned capability awareness: point the Skills pattern at the model's knowledge of its own host
- [FEATURE] Add a 'propagate-the-fix' (downstream consistency propagation) pattern to the Workflow tool's quality-patterns library
- [FEATURE] Sort or alphabetize artifacts in the chat/project artifact panel (Desktop, macOS)
- [Feature Request] Add a setting to hide the weekly usage/rate-limit indicator above the prompt input
- [Feature Request] Add config option to always preserve user message on Esc cancel
- [Feature Request] Add option to disable or relax Safeguard restrictions for local development
- [FEATURE] Configurable suggestions for denied commands
- [FEATURE] Allow to resume interrupted generation without creating a new conversation turn
- [Feature Request] Support for legacy Lisp machine emulator and custom OS interfaces
- [Feature Request] Support for broad project use cases in automation workflows
- [FEATURE] Add Dependabot alerts (vulnerability_alerts) read permission to the Claude GitHub App
- Feature request: default-on Auto-fix for PRs opened by scheduled Routines
- [FEATURE] Arabic text appearance
- [Feature Request] Document backwards compatibility requirements for CDN migration
- [Feature Request] Add team member Claude agent delegation for unavailable users
- [FEATURE] Side chat (read-only auxiliary window) can't apply edits to the main conversation directly
- [FEATURE] Configure a subagent's model and effort from settings, without forking its prompt (plugin-shipped agents)
- [FEATURE] Add worktree directories to Claude session search on VS Code open
- [FEATURE] Trigger a local routine on demand via a local command / IPC (not only cron)
- [Feature Request] Expose bindable actions for permission-dialog options and remap digit quick-selects
- [FEATURE] Desktop app: allow moving a Code session into a Claude Project
- [Feature Request] Enable direct in-chat editing of partial file outputs without requiring external editor
- [FEATURE] Claude Desktop: opt-in setting to auto-approve write/modifying MCP tools per server
- [FEATURE] Clear oversized inputs of validation-failed tool calls from context — a failed Write's full content is dead weight
- [Feature Request] Add option to manage routines created by Claude Tag in a Slack channel
- [FEATURE] Add Multilingual (i18n) Support for Slash Command Descriptions (/ 슬래시 커맨드의 자동완성 때 뜨는 설명문구를 다국어 지원하는 방향 제안)
- [Feature Request] Add option to keep file tabs open after accepting edits
- [FEATURE] Dispatch should support AskUserQuestion button UI for better mobile UX
- [FEATURE] More Projects on the Free plan
- [FEATURE] Add Zig syntax highlighting for code blocks (ready-made grammar: highlightjs-zig)
- [FEATURE] Support fork agents in Workflow scripts
- Feature suggestion: built-in "read response aloud" (TTS) mode
- [FEATURE] Routines have no durable in-app notification surface — findings only survive by being written into Calendar/Gmail
- [Feature Request] Allow claude-code to work on fantasy/fictional biology-theme content projects
- [FEATURE] Forward URL fragments into published artifacts so section deep links work
- Option to paste an image file path as literal text (don't auto-attach)
- Feature request: Proactively suggest switching to a higher-tier model (e.g., Fable) based on task complexity
- Session-wide model switch to Fable billed all turns at premium rate — no warning when scoping was intended to be per-task
- [FEATURE] retrieve the search index's cached/extracted copy of a page
- Keybindings: direct model-select and effort-adjust actions (not just open-picker)
- [FEATURE] Sort File Suggestions by Creation or Updated Date
- Agents view: @ project picker is prefix-only, so interior substring matches never surface
- [FEATURE] Add a setting to exit with a single Ctrl+D instead of pressing it twice
- [FEATURE] Allow Ctrl+D, Ctrl+D to exit `claude agents` (consistent with `claude --resume`)
- [FEATURE] Interactive session deletion in the `--resume` picker (like Ctrl+X in `claude agents`)
- Bash tool shell runs with `expand_aliases` enabled — shell aliases can silently rewrite commands after PreToolUse hook approval
- [FEATURE] VSCode extension: render images inline in the chat panel instead of the "[Image]" placeholder
- [FEATURE] prompt for `cleanupPeriodDays` during initial setup
- [FEATURE] Ability to revoke Claude's access to a specific tab without closing/reloading it
- I'm not able to generate an issue title from that input as it doesn't contain a bug report or feature request description. Please provide details about the actual problem you're experiencing with Claude Code, including: - What you were trying to do - What
- [BUG] No safeguard before an agent bulk-creates dozens of PRs against a real external repo (91 PRs auto-closed + repo locked me out)
- [FEATURE] Configure list of session history to display in Claude desktop
- [FEATURE] /export to export sessions in .html file
- [FEATURE] /nudge — send an ephemeral one-turn instruction that doesn't persist in conversation context
- [FEATURE] Chips with dropdown functionality e.g. workfolder should have a chevron or other indication that they are dropdown
- [FEATURE] Permission Dialog Position Conflicts with Windows Notifications
- [FEATURE] VS Code extension: render MCP `claude/channel` push notifications in the native panel (parity with the terminal TUI)
- [Feature Request] Restore discontinued Fable model support
- [Feature Request] Official support for Claude Desktop on Arch Linux and derivatives