Digest / September 16, 2026
Claude Code Issues Digest — September 16, 2026
269 new issues filed ·
188 resolved in anthropics/claude-code (UTC day).
New issues (269)
- Feature request: built-in, cross-platform way to list running Claude Code sessions and their state
- [Feature Request] Review content moderation for legitimate security education use cases
- VS Code extension: Claude Code occupies two tabs (chat + account/usage panel), unlike other AI extensions
- [Bug] Model switched without user request or clear reason
- Cross-session and mid-turn messages silently dropped (SendMessage reports success, message never arrives)
- Feature request: multi-select quote reply (batch quotes into one reply)
- SONNET 4.6
- [Bug] Claude frequently fails on simple confirmation and lookup tasks despite having necessary tools
- Desktop app update wiped local scheduled tasks and session list (claude.ai web still shows them)
- [BUG] Context isolation failure
- [BUG] MCP OAuth: DCR client_name "Claude Code (<server>)" is rejected by servers that validate RFC 7591 metadata strictly (Calendly MCP → 400 invalid_client_metadata); recurrence of #59445 / #73293, still present in 2.1.271
- I don't see a specific bug report or technical issue in your message. This appears to be a general statement about improving website security rather than a bug report for Claude Code. If you have encountered a specific bug or issue with Claude Code while
- [Bug] pkill -f process matching kills unintended user processes on occupied ports
- Merge/push-to-main classifier blocks git push and gh pr merge even with full permissions.allow and Auto Mode configured
- [BUG] Security-guidance plugin hook file missing from session snapshot, causes repeated "No such file or directory" notifications
- [BUG] Desktop Code tab: enabled ~/.claude plugins never load into sessions — regression 2.1.237 → 2.1.260, still broken in 2.1.270
- Desktop app: file links in replies break silently on moved files, absolute paths, and spaces
- [BUG] Desktop (Windows): a forced re-authentication mid-session silently drops the prompt just submitted in another session
- [BUG] /model picker omits Fable 5.1 on personal Max 5x account (non-org) despite entitlement
- Desktop-app-installed claude.app bundle has no published checksum and carries entitlements the bare CLI does not
- [BUG] Cowork local sandbox: proxy returns 403 'Connection blocked by network allowlist' for cdn.playwright.dev although it is in the session's default egressAllowedDomains
- [BUG] archive_session with session_id "self" hangs for hours in scheduled runs, delaying other scheduled tasks
- [BUG] Scheduled task runs report "succeeded" when they were aborted at a permission prompt or produced no output
- [BUG] create_scheduled_task has no permission-mode parameter, so MCP-created tasks always stall unattended
- [BUG] GitHubでバグ報告
- [Bug] Agent ignores user instructions and context
- [Bug] Legitimate network data flagged as suspicious content during report generation
- [FEATURE] For Claude Sonnet, add an option in settings for system time at the end of every prompt and at the beginning of every response. NOTE: There have been other requests for timestamps in the chat.
- VS Code extension 2.1.272: chat file links with non-ASCII (Korean) path components no longer open (regression from 2.1.270)
- Remote Control phone voice channel: uncommon proper names transcribed inconsistently (e.g. "Jarvis" heard as "Josh," "Travis")
- [Bug] Claude Code agent fabricates data field significance and overclaims fix adequacy without verification
- [BUG] auto mode adds a fixed ~12.3s to Bash commands that skip the permission-rule fast path (controlled test: 12.32s vs 0.15s for the identical command)
- [BUG] Classifier fallback (model_refusal_fallback / apiRefusalCategory="cyber") silently swaps model mid-session; substituted usage billed as regular subscription usage
- HubSpot connector manage_crm_objects tool: createRequest rejected as String instead of JSON object
- [BUG][Windows] Desktop git-shadow probes leak orphaned git.exe on large SMB repos — timeout kills the Git for Windows launcher, not git; 55 procs / 20 GB in 1 h (follow-up to #86646)
- [FEATURE] Let a subagent opt out of the advisor (per-agent key or settings scope)
- [BUG] Auto mode tells the model to edit files via Bash, which silently disables checkpointing / `/rewind` (no Restore code on any prompt)
- [Feature Request] Add notes field before submission in questionnaire workflow
- [BUG] Claude.ai Projects: project files are not injected into the sandbox when a chat joins a project after the container has already been created
- [Bug] Agent executes tasks without user approval
- [MODEL]
- Too much detail on PRs
- [FEATURE] Desktop Files pane: option to show dot-prefixed files and directories (hidden files)
- In-app Browser preview reports "not enabled on this install" despite setting being ON
- [Feature Request] Add intermediate context window model option between 200K and 1M tokens
- code-review plugin: parallel review agents run in background and are abandoned → silent review with is_error:false
- Automatic extension resolution fails resolving require('dotenv') — tries to open src/dotenv
- [BUG] Mouse drag-select copy silently fails on WSL2 in the VS Code integrated terminal: highlight shows, no clipboard write, no notice (2.1.273; Windows native works)
- [BUG] Japanese characters in custom title cause malformed JSON, rename doesn't persist
- [BUG] VS Code extension: settings writes fail with EPERM because the Windows profile path is character-reversed for RTL (Hebrew) usernames
- [FEATURE] Two-phase slash command — context-aware planning subagent feeding a clean-context async execution subagent, off the main agent's critical path
- [WHINE] Persistent monitor configuration removed unexpectedly
- [BUG] Code and Settings tabs missing on Linux — GrowthBook feature flags reload as 347/347 changed amid repeated "account change" refreshes
- [BUG] Adobe for Creativity plugin: OAuth sign-in always fails (gateway 403s break discovery; DCR tokens rejected)
- UserPromptSubmit fires for agent/system-injected messages with no prompt_source/is_meta in the payload — hooks cannot tell them from typed input (prompt-injection surface)
- [FEATURE] Setting to opt out of markdown rendering for my own prompts
- [FEATURE] /compact <args>: the args are prompt-shaped but never run — session goes idle while the raw <command-args> block stays inert in the new context
- [FEATURE] Desktop app: remember additional folders per project across sessions
- [BUG] A `userConfig` field with `options` prevents plugin installation
- [Retracted] Windows commit exhaustion was caused by a runaway test in my own code, not Claude Code
- [Bug] Remote Control feature unavailable on Pro plan
- VS Code panel silently drops Auto from the permission-mode picker while the extension host reports canEnterAuto=true
- macOS: auto-update triggers repeated, unlabeled "<version> would like to access data from other apps" TCC prompt
- [Model Behavior] Claude Science (Windows) Safety classifier cascade degrades research to Haiku 4.5, which then hallucinates, destroying valid scientific work
- [FEATURE] /compact <args>: the args are prompt-shaped but never run — session goes idle while the raw <command-args> block stays inert in the new context
- Feature request: historical usage-limit exhaustion count + timeline for Max/Pro accounts
- macOS: voice dictation dies in every running session after the auto-updater replaces claude.exe ("No audio detected from microphone"; TCC denies the old PID)
- "notification push Gmail qui cesse silencieusement après ~7 jours, trigger type Email/API".
- Desktop app: @-mention renders literal "@[SessionName]" bracket text on first insert, needs delete+reselect to show as chip/link
- Bash deny-rule denial doesn't name the matching rule or its settings scope — server-managed deny rules are invisible to the developer they apply to
- [BUG] Windows: desktop updater overrides its own "Claude is working" deferral and force-restarts mid-task, destroying a running unattended Claude Code job
- [Bug] Opus 5 fails to verify assumptions before proceeding with tasks
- [Bug] Anthropic API Error: Model refuses request with vague "bio" classification
- "Approaching weekly usage limit" banner fires at 54% weekly usage
- Feature request: Czech (cs) localization for Claude Code desktop UI
- [Bug] Anthropic API Error: Safeguards incorrectly flagging legitimate coding and security tasks
- [FEATURE] Let a session start another session programmatically, visible in the sidebar
- Desktop app: "Group by project" sidebar rows cannot be archived or hidden
- [MODEL] Opus 5: publishes conclusions whose own validation check failed; disconfirming evidence downgraded to caveat
- Sidebar shows no sessions after update — local data intact but not loaded
- [BUG] Cowork and Code: No working IDE endpoint available - Desktop app cannot discover IDE plugin endpoint
- [BUG] Duplicate "security-guidance" hook-failure notification loops indefinitely across sessions
- [Bug] Claude Code executes actions without explicit user authorization and cannot be reliably stopped mid-execution
- Dangerous-rm approval dialog displays a target resolved against the session cwd, not the cd target
- [Bug] Weekly usage quota suddenly decreased by 31% without explanation
- [Bug] Opus 5 cyber safeguard incorrectly blocks defensive security review of user's own code
- Desktop app: file links outside the working directory render clickable but dead-end in 'Couldn't find this file'
- OAuth token exchange fails with HTTP 400 on fresh /login and setup-token (VS Code Remote Tunnel to Linux)
- Skill body is rewritten when arguments are passed: $1/$2 inside SKILL.md get replaced by words from the argument string
- Desktop app: no user-level way to disable the "How is Claude doing this session?" survey
- Desktop app: show full filename on hover for truncated file attachment cards
- [Bug] Opus 5 produces hollow, non-factual French prose despite in-session corrections
- [BUG] Claude Desktop completion chime cannot be disabled, plays repeatedly, no app setting exists
- [FEATURE] Pinned inline threads: ask about a selected part of Claude's reply without derailing the conversation
- Trailers in a -F commit message duplicated once; not reproducible
- VS Code panel hardcodes the plan-approval buttons to acceptEdits/default, so the CLI's "Yes, and use auto mode" row never appears
- [Bug] Claude Code making unverified claims and false positive issue detection
- MCP client rejects tool calls omitting optional parameters; deadlocks on tools with mutually-exclusive parameter groups
- [Bug] Anthropic API Error: Request ID req_011Cf6pmWJxTSs5xQwLUp4Qk - Safety Guidelines Violation
- (withdrawn)
- [FEATURE] Desktop app: full session list ("Show all") doesn't display or allow managing custom group assignment
- [Desktop] Inter-session messages now render as a collapsed "Received message from …" row — please add a setting to show them expanded inline
- Desktop: new chat should default to the folder selected in the sidebar
- [BUG] Archiving a project doesn't archive its child sessions; manual archive afterward fails, sessions still show under active filter
- [BUG] Attribute 'effort' is not supported by VS Code skills. Supported: argument-hint, compatibility, context, description, disable-model-invocation, license, metadata, name, user-invocable.
- [MODEL] Auto-mode credential classifier blocks a previously-working SQL connection with no config change
- [Feature Request] Support for Claude Opus 5 model selection
- [BUG] Resuming a background subagent misses its prompt cache: messages_changed, no thinking blocks
- Hook quebrado do plugin security-guidance gera loop de notificações automáticas idênticas (~30+ repetições)
- [MODEL] Opus 5 and Sonnet 5- Behaving unintelligently
- Windows: desktop app force-terminates active sessions — launcher passes ForceTargetApplicationShutdownOption in a repeating repair-register loop
- Desktop app: message input box refills with previously sent message after every send
- [Feature Request] Improve agent task orchestration for complex multi-step workflows
- Sessions archive/drop unexpectedly and scheduled tasks don't sync to iOS remote control
- [Bug] Model loses context of coding specification mid-session
- [BUG] edited reply is locally reverted after login
- VS Code panel: a skill invocation with a long argument is rendered in full and covers the whole panel
- Auto mode: a PreToolUse hook "allow" should resolve the action like a permissions.allow rule
- [BUG] stream-json: result event omits the `result` field entirely after a control_request interrupt
- [BUG] allowedTools `Bash(printf *)` denies `printf x > file` with no message that output redirection defeated the prefix rule
- VS Code: after an extension host restart, Claude panels never re-attach, so every session in the window looks frozen until Reload Window
- [Bug] Diff view not displaying when agent operates in Git worktree
- [BUG] I lost all the conversation sessions I had.
- Hosted connector drops `blob` from embedded resources: -32602 invalid_union for any binary document (same payload accepted by the engine client)
- Keep getting flagged wrongly for [cyber]
- fire_trigger spawns a new blank session instead of resuming the bound persistent session
- Windows/Git Bash: the Bash tool dies for the rest of the session (unexpected EOF at a line past the command / ENAMETOOLONG uv_spawn)
- VSCode extension: session shows "Chat processing error" then resets to blank welcome screen after Cmd+Q relaunch
- Strg+Enter ("Für später vormerken") sendet Nachricht sofort statt sie in die Warteschlange zu stellen
- Chat mode vs. Code (</>) mode toggle is too small/low-contrast, easy to switch by mistake
- [MODEL] Compiler errors incorrectly treated as satisfying TDD RED gate
- [Bug] Claude 3 Opus ignores agent constraints and executes unintended actions
- [BUG] macOS dictation (voice input) is misrecognized as English in the chat input, even when macOS dictation language is set to Japanese only
- [Bug] Anthropic API Error: Safeguard rejection on legitimate code fix requests
- computer tool's screenshot/find/read_page/scroll consistently timeout with "document_idle" wait failure across ALL pages
- Claude Desktop 3p + custom gateway (`ANTHROPIC_BASE_URL`): sandbox network pinned to gateway + api.anthropic.com, `sandbox.network.allowedDomains` ignored even after full app restart
- Google Calendar connector: list_events/search_events return calendar metadata instead of events
- [Bug] Lagging performance during reasoning extraction
- [BUG] Claude silently works around blockers instead of surfacing them — user has no hook to intervene
- [BUG] Skills declared inline in `marketplace.json` under `strict: false` are not registered (2.1.112 works, 2.1.270 broken)
- [Bug] Anthropic API Error: Overly restrictive safeguards blocking legitimate cybersecurity tasks
- claude-in-chrome: account switch (personal Max → Teams) breaks connection; reinstall fixes
- [Bug] Tool results leaking between sessions in multi-session environments
- [Bug] Tool execution timeout too short in check_all.sh (10 min limit)
- iOS Simulator panel: claude-ios-sim crashes (SIGABRT) on attach via CoreImage Metal binary archive
- [FEATURE] Fleetview, control project/group sort order (alphabetical / pin to top)
- [Bug] Unexpected CJK characters in Claude Code output
- [FEATURE] Budget-aware subagent orchestration: premium-model fan-out burned ~70% of a weekly Fable 5.1 allowance, mostly on subagents killed by the rate limit
- Stop hook reports phantom "unpushed commit(s)" after a squash merge deletes the remote branch
- I don't have enough information to generate a GitHub issue title for Claude Code based on your message. Your report appears to be about:- A bot attack on a load balancer- Backend API exposure- A request to have "Fable"
- [Bug] Generated examples lack logical coherence with prior context
- [FEATURE] Fleetview, allow renaming of projects
- Opus 5 reasoning_extraction false-positive blocks benign messages and built-in workflow-authoring skill docs
- [BUG] Initial agent-type listing never emitted on turn 1 of a session created by plan approval with "clear context"
- [Bug] Incorrect content moderation classification: legitimate questions marked as cyber
- [BUG] All kinds of problems during last night's session
- Cloud routine (CCR) sessions permanently hang at startup, 100% reproducible over a month
- Remote Control: archiving from the app disconnects the bridge (code 4090) but leaves the local bg session, registry entry and job record alive; app then shows it still open
- [BUG] Token Usage count
- I don't see a specific bug report or technical issue description in your message. To generate a proper GitHub issue title, I need details about: - What specific problem or error occurred - What the expected behavior should be - Any error messages or stack
- [Bug] API response timeout delays every 5 minutes
- [Bug] Model switches unexpectedly during active development session
- [BUG] Model generated a fake hostile "user message" and replied to it within a single response
- [Feature Request] Add conversation linking/anchoring to specific messages
- [Bug] Desktop app Claude Code tab: inline \( ... \) math silently fails above 58 characters
- [BUG] CLI ExitPlanMode prefers auto over the configured acceptEdits default — plan-exit can't return to acceptEdits
- code-review plugin ends its turn while waiting on subagents in one-shot (-p) runs, posting nothing
- Title: Show exact dates for sessions in Code tab sidebar, not just Today/Yesterday/Older Body:
- VSCode extension: no hook fires when user rejects a tool permission prompt
- [Bug] Claude Code CLI output not visible in mobile Claude app
- [BUG] German output: the model coins compound nouns that do not exist and uses real words in meanings they do not have; corrections do not survive the turn, the session, or a written rule (Opus, Fable, Sonnet)
- [BUG] external file changes are not notified to the agent before an Edit tool call when the file was only Read (not previously Edited) — no “modified since read” guard, no change notification
- [Bug] False Positive Content Filtering Blocks Legitimate Input
- Memory files become unopenable in file-preview pane after session's working directory changes
- [FEATURE] Add more color options to `/color` command
- [BUG] Merged Home: chat threads and Cowork sessions coexist in one project with different capabilities and no visible mode indicator
- Invalid tool_use name (>200 chars) persisted to transcript poisons the session with 400 on every later request
- [Bug] Claude refuses to perform web application login for UI/UX testing, logging into a form is a line it won't cross
- Wrong titel
- [Bug] Missing usage metrics and context progress bar display
- [Bug] Plugin-sourced MCP servers silently ignored under `allowManagedMcpServersOnly` enterprise policy despite `serverName` allowlist match
- prompt=consent hardcoded on every MCP OAuth authorize request breaks Entra tenants with user consent disabled — affects Claude Code, claude.ai web connector and desktop alike (#49722 was closed without a fix)
- [Feature Request] Allow security testing prompts for approved penetration tests on local systems
- Cowork desktop: MCP elicitation (2026-07-28 MRTR) dropped via print-mode branch, tool call hangs 180s
- [BUG] Sidebar sessions/projects disappeared from Code tab, data still exists
- [BUG] Interactive → background session handoff still leaves title-only .jsonl stubs on 2.1.270–2.1.273 — 7/7 stubs traced to `bg spawned|claimed-spare <id> (slash)` (follow-up to #81662, #85404)
- /clear in the desktop app does not leave the previous session recoverable (unlike CLI)
- [Bug] Claude Code hangs during auto-compact operation
- [FEATURE] Expose session readiness to orchestrators: error class on StopFailure and a degraded state in ListAgents
- [FEATURE] SessionIdle hook event with exit-2 continuation for any session (generalizes TeammateIdle; reopens #68060)
- [BUG] Bare skill-only slash command as first message in a new session sometimes hangs indefinitely (never reaches "Thinking...")
- [FEATURE] Add utm_source=claude-code to all urls
- [BUG] Prompt cache misses on every background session: per-session job-dir path contaminates the `cacheScope:"org"` system-prompt block
- I'd be happy to help generate a GitHub issue title, but I need more information about the bug report. Could you please provide the details of the issue you'd like to report? This should include: - What were you trying to do? - What happened instead? - Any
- Session tab rename is overwritten by a stale webview title; rename only appears after restarting VS Code
- [FEATURE] Highlighted Scripting
- [BUG] Cowork task starts with no device_* tools while the bridge is authenticated (Windows 10 22H2, after KB5129236)
- Background Bash task SIGTERMed by its own supervising CLI process on Linux — introduced in 2.1.246 (2.1.245 clean)
- [BUG] Claude code panel not opening up in bottom terminal panel.
- [BUG] VS Code extension sessions don't show up in the Desktop app
- [BUG]
- HTTP MCP OAuth: pending flow state lost across a turn boundary, breaks manual complete_authentication on remote sessions
- [Bug] Claude repetitively outputs "privately" and "what I need next" messages
- [Bug] Agent fails to report output file location and writes to inaccessible directories
- Model selection has been unreliable since 2026-04-28, and the issue naming the mechanism (#73881) has been open 75 days — users are still billed at a tier they explicitly left
- Cross-session message delivery stopped surfacing in the recipient session — regression in claude-desktop ≥ 1.46388.2
- [BUG] Gmail connector: all tool calls fail with "Precondition check failed" even after full OAuth reauthorization
- Desktop built-in browser can't grant standing permissions for .local hosts (WordPress Studio sites)
- [BUG] Desktop split view: unsent drafts disappear when two sessions swap panes and only come back in the original pane
- [BUG] /context reports 0 tokens for MCP server instructions that cost 653 per turn
- [BUG] Desktop app "Can't reach your computer" — persists after ruling out every local network/OS cause
- Gmail MCP: Gmail MCP API confirmed enabled, still "caller does not have permission" (follow-up to #77210)
- [BUG] "Word (By Anthropic)" extension: replace_text always fails with AppleScript syntax error
- [BUG] --teleport of a web session restores only the first turn; later turns missing
- [FEATURE] Identify Claude Desktop Instances in Tray Icon Tooltips
- [BUG] MacBook был offline в VPN → переподключил Wi-Fi → появился в сети → но сама сессия Claude Code не восстанавливается (перезапуск программы, полный выход/вход — не помогли).
- [Feature Request] Improve codebase scanning accuracy and simplify technical explanations
- [Bug] Build script export path fails on repeated configuration changes from A to B
- Atlassian connector widgets render with a white background in dark mode and are always expanded inline
- [Bug] Built-in skills not recognized as installed
- Claude Code rewards user escalation: calm requests get partial work, angry ones get the real answer
- [BUG] TUI status line doesn't update permission mode after /config change (fullscreen, macOS)
- [Bug] Remote Control messages stall when status panel (QR code) is open in terminal
- [BUG] skillCreationEnabled not being enforced in the Desktop App
- Desktop app: Run button on a bash code block always fails with "Failed to send the command result to Claude"
- [Bug] Excessive output verbosity
- [FEATURE] Disclose transcript scope and retention in the /bug and /feedback dialog before submission
- [Bug] Claude Code ignoring user instructions and modifying approved requests
- [Bug] Artifact publish/watch events no longer trigger session wake
- This is not a bug report, so no GitHub issue title is needed. You've identified that Claude Code correctly couldn't access folders on a separate machine—this is expected behavior and will be resolved once you establish the necessary access.
- [BUG] Code tab forces full re-sign-in after a brief network switch (hotspot to Wi-Fi), killing the active session
- [BUG] macOS: claude auth login reports success but CLI stays unauthenticated - hasAvailableSubscription false on an active Stripe subscription, and no credential is persisted
- [DOCS] plugins-reference still shows a flat `agents/` tree and omits recursive scanning and path-scoped identifiers
- [FEATURE] /context in the classic renderer should default to the compact summary, with /context all for the full listing
- [BUG] opening temporary chat in cc desktop app cleverly switches ur model for more consumption
- [Bug] Command history bleeding across simultaneous Claude Code sessions
- [Bug] Claude Code consuming excessive tokens and failing to complete basic file operations
- [Feature Request] Add session observation/monitoring mode for improved collaboration
- [Bug] Unable to determine issue - insufficient information provided
- iOS Simulator panel freezes at FPS:0 and only a full app restart recovers it
- [Bug] OAuth token deleted or account disabled after form submission with issues
- [BUG] No environment scoping on the shared ~/.claude state root: consent, config schema, transcript retention and permission defaults couple CLI and desktop sessions
- You've hit your weekly limit · resets Sep 18 at 3am (America/Los_Angeles)
- [BUG] device_bash / Cowork local device bridge fails on Windows 10 (10.0.19045) — same symptom as #94266/#92958, but NOT Windows 11
- [BUG] Cowork Windows Insider Beta 26220.9472: KB5129195 Plan9 fix is not applicable to this branch, device_bash still dead
- [Bug] Safeguard triggered on session start with benign cyber work context
- Remote Control failed to connect: initialization failed
- Subagent spawn records carry no status or lifecycle timestamps, so orphaned background agents display as "running" indefinitely — including across a reboot
- [BUG] AltGr characters are lost in Windows Terminal since the kitty keyboard protocol is enabled unconditionally
- [Bug] Fable requests incorrectly marked as "reasoning_extraction" regardless of content
- [MODEL] Sonnet 5 (Claude Code returns "Sonnet 5 can't help with this" when reading or receiving a valid USPTO API key credential)
- [Bug] Anthropic API Error: 500 Internal Server Error
- [BUG] [VS Code] Plan mode indicator stays selected but subsequent messages run under "Edit automatically" permissions
- [Bug] macOS TCC permissions lost on auto-update due to versioned binary paths
- [BUG] Changing tui in settings.json blanks running fullscreen sessions not started with CLAUDE_CODE_NO_FLICKER
- Claude for Chrome: site-level permissions unavailable for some domains, with no way to grant them
- Three days, most of a week's Claude usage, zero trades through Step 4: reviewed code never assembled or run, agents spun on data perfection, and the PM escalated to the top model tier against written instructions
- [Bug] Repeated macOS file access permission prompts on each invocation
- [Bug] Excel file reconstruction from screenshots triggers false security flags
- [BUG] Linux: login dead-ends at "Finish sign-in in the Claude app"
- I don't have a bug report to analyze in your message. Could you please provide the actual bug report or issue description you'd like me to create a GitHub issue title for? Please share: - What error or problem occurred - What you were trying to do - Any e
- For three blissful hours I actually believed the Claude app had evolved into something that gets shit done — the board was green, the code had never run
- [BUG] Desktop SSH session + Remote Control: while the laptop sleeps, claude.ai connectors are unusable from the phone because they execute inside the sleeping Desktop app — answering the PreToolUse permission hook is not enough
- [BUG] Chat tab strip (multi-session view) missing in VS Code extension after update to 2.1.273
- [FEATURE] Agent view: /new panel to set model, effort, name, permission mode and isolation before dispatching a session
- MCP tool schema loses type/validation info between server and Claude's tool-calling path
- [FEATURE] Vim mode: change cursor appearance based on insert/normal mode
- [FEATURE] Let /branch register the new session in agent view without a separate /bg
- effortLevel setting cannot persist the "Max" effort tier the UI offers
Resolved issues (188)
- [Bug] Anthropic API Error: Overly aggressive safety filtering on legitimate system signal operations
- [Bug] Context indicator appears on unselected files, causing incorrect context in thread
- update-config skill generates Write() rules instead of Edit() rules
- [FEATURE] /compact <args>: the args are prompt-shaped but never run — session goes idle while the raw <command-args> block stays inert in the new context
- [Retracted] Windows commit exhaustion was caused by a runaway test in my own code, not Claude Code
- (withdrawn)
- [BUG] Attribute 'effort' is not supported by VS Code skills. Supported: argument-hint, compatibility, context, description, disable-model-invocation, license, metadata, name, user-invocable.
- Mobile web (claude.ai): long-press opens the message menu instead of selecting text — no way to copy a partial selection
- [Bug] Tool Safeguard Incorrectly Flags IPv6 Modem Configuration Tool
- [Bug] Summary output displays out-of-order messages causing false alarm alerts
- [BUG] "Vercel MCP integration returns 403 Forbidden on deploy_to_vercel for existing/manually-created projects, works only once on brand-new project names."
- [Bug] Claude Code hangs on long-running command execution causing excessive delays
- [FEATURE] Desktop app: full session list ("Show all") doesn't display or allow managing custom group assignment
- [FEATURE] Always-visible active provider/account indicator in Claude Code CLI and VS Code Extension, with per-provider color theming
- [Feature Request] Emit OSC-8 hyperlinks for file paths in CLI output
- [Bug] Potential session/cache leakage between workspace instances or consumer accounts
- Session auto-title latches onto an interrupted first message and never regenerates
- Weekend post-mortem: fuck-all got built, and Fable still ate its usage on process it invented instead of the work I asked for
- Background Bash tasks get lost/duplicated, turning a simple deploy into 10+ hours of retries
- Full-width digits (１-９) don't select options in the question dialog
- [FEATURE] User control over prompt suggestions under usage limits (keep available at allowed_warning, optional lighter model)
- Status indicator dots for sessions are hard to distinguish (running / awaiting input / unread) — improve contrast or make configurable
- [FEATURE] Add opaque/deep compaction for long-running coding sessions
- Expose per-message timestamps to Claude's context (Cowork + Chat gap — no workaround available)
- [BUG] Archiving a project doesn't archive its child sessions; manual archive afterward fails, sessions still show under active filter
- [FEATURE] Stable, documented schema for ~/.claude/projects/<encoded-cwd>/<uuid>.jsonl line types
- [desktop] Add /reload-plugins command for in-process plugin refresh (parity with CLI)
- [FEATURE] add optional timezone field (IANA tz, e.g. America/Los_Angeles) to scheduled-trigger cron expressions
- Cost estimation blind spots when building agent systems with Claude Code
- [BUG] Hint/placeholder text uses SGR 2 (faint), which renders at full brightness on terminals without dim support (e.g. mosh)
- [Feature Request] Remove safeguard gating for agent harness and memory project implementation
- [FEATURE] Surface workflows still running in another session for the same project
- [FEATURE] Defer MCP server instructions on demand, the way ToolSearch already defers tool schemas
- [FEATURE] Building in natural language is amazing — the file layer just hasn't caught up yet
- I appreciate you reaching out, but I'm unable to generate a GitHub issue title from this message because it doesn't contain a clear technical bug report or feature request related to Claude Code functionality.
- [FEATURE] Desktop app "Open in" menu: support JetBrains IDEs (Rider) / custom editors
- [Feature Request] Support filtered/partial content generation for course indexes
- [Feature Request] Add support for Claude 3.5 Opus model variant
- Ok [Feature Request] Whitelist or exempt epidemiological modeling code from safety filtering
- [FEATURE] Let Pro/Max subscription usage count toward API key requests, so subscribers can use their plan in third-party tools without needing separate token billing
- [FEATURE] Reusable saved prompts for /btw (side chat) — and a /compact that drafts its own instructions there
- [FEATURE] CLI: `code` is the only block-level token given neither a container nor a screen-reader form
- Feature request: server-injected per-installation secret for key wrapping (software "TPM")
- [FEATURE] Add SubTasks in TaskCreate
- Feature request: attach an existing session as a teammate (cross-session messaging)
- QuickBooks Online connector has no support for Classes anywhere in its toolset
- QuickBooks Online connector missing time-activity/timesheet creation tool
- Feature request: periodic spinner verb rotation during long-running operations
- [FEATURE] Named model presets: multiple /model picker entries with preset effort levels
- Feature request: visual nudge mode — select an on-screen element and arrow-key its position
- [FEATURE] VS Code extension: retrieve sessions from a different projects/windows
- [FEATURE] Claude in Chrome: pin the MCP tab group to a designated Chrome window
- [FEATURE] Session timer that excludes idle time to determine how long actual work took in a session.
- Feature request: allow custom ordering / pinning in the "/" slash-command autocomplete menu
- [FEATURE] Support /desktop session handoff on Windows ARM64
- [FEATURE] Render the session's statusLine in the Remote Control web/mobile UI
- [FEATURE] Desktop app: make it obvious how to collapse an expanded thinking block (and hold scroll position)
- [Feature Request] Support for security tool modification workflows in authorized red team contexts
- [Feature Request] Apply unused Pro plan credit to Max plan billing
- [Feature Request] Add support for Claude 3.5 Haiku model option
- [Feature Request] Extend session timeout or add session persistence for long-running security audit tasks
- [Feature Request] Support for security testing and penetration testing workflows
- google-workspace: get_events should expose colorId field when detailed=true
- [Feature Request] Add support for authorized security testing flag or context in API calls
- [FEATURE] Allow enabling a channel on an already-running session (--channels is start-time only)
- [Feature Request] Model switching safeguards too aggressive, blocks legitimate use cases
- [FEATURE] Support folder-based project trees: auto-create + group chats per subfolder
- [Feature Request] Long-running tasks need persistent state and decision guardrails to prevent session abandonment and course reversal
- [Feature Request] Improve fable API guardrail error messaging and documentation
- [Feature Request] Add safety mechanisms to prevent silent data corruption from model changes
- Feature request: Make it visible whether a session runs locally or remotely (session list indicator)
- [FEATURE] Become a LinkedIn "Connected Apps" partner to showcase Claude Code usage on profiles
- [Feature Request] Add more color options to `/color` command
- iOS App Account Switching for Remote Control
- [FEATURE] Render thinking blocks in Remote Control sessions on claude.ai/code and the mobile app
- Per-project visual identity in Cowork: accent color, icon, or cover image
- [Feature Request] Improve Claude 5 Sonnet instruction following and code quality
- [FEATURE] Routines API: expose GitHub/API triggers and delete, so a routine can be managed without the web UI
- [FEATURE] Ask the user earlier on long-running tasks: 4 of 5 wrong conclusions were caught only by external interrupts
- [FEATURE] Browser pane microphone-block banner reappears and blocks page interaction on every getUserMedia() call
- [Feature Request] Add ability for advisor agent to force resume failed agent processes
- [FEATURE] Claude is too aggressive about screen space management; give me way to turn that off or at least down
- [Feature Request] Mirror input-editor kills (Ctrl+K, Ctrl+U, Ctrl+W) to the system clipboard, or expose an opt-in setting
- I need clarification on your issue to generate a proper GitHub issue title. Could you please describe: 1. What problem are you experiencing with Claude Code? 2. What is the error message or unexpected behavior? 3. What were you trying to do when this occu
- FUCK OPUS 5 HOLY SHIT
- Wrong titel
- [BUG] Desktop app rejects MCP calls that omit optional parameters (expected, nonoptional, rerecived undefined)
- Claude Desktop tray icons are indistinguishable when multiple instances are running
- [BUG] Gmail connector: all tool calls fail with "Precondition check failed" even after full OAuth reauthorization
- [FEATURE] Let a session start another session programmatically, visible in the sidebar
- iOS Simulator panel freezes at FPS:0 and only a full app restart recovers it
- [BUG]
- [Bug] Agent fails on basic reasoning tasks that should be trivial for Claude 3.5 Sonnet
- [Bug] Safeguards false positive: Constant triggering on benign messages
- [Bug] Claude Code deleted files unexpectedly during hallucination
- Send an email regarding security issues of a service provider I'm using for a client
- [Bug] Anthropic API Error: Rate limit or quota exceeded on standard message
- I don't see any bug reports in your message. Please provide the specific bugs or error messages you'd like me to help create GitHub issue titles for, and I'll generate appropriate technical titles for each one.
- [Bug] Invalid model name anthropic/claude-3.5-sonnet persists after reset and reauth
- I don't have a bug report to analyze. The input "ssdf" doesn't contain any meaningful issue information. Please provide a detailed bug report or feature request describing: - What happened - What you expected to happen - Error messages or logs - Steps to
- [Feature Request] Add scrollbar to fullscreen output view for faster navigation
- [FEATURE] Interrupt blocking TaskOutput when the main agent receives a SendMessage
- [Feature Request] Review safety guardrails sensitivity for non-harmful workflows333
- [FEATURE] Optional "Are you sure?" confirmation when closing the desktop app / exiting while the agent is working
- [FEATURE] Cross-session "needs your attention" inbox — questions, decisions, and findings aggregated outside the transcripts
- [FEATURE] Auto-archive on PR close: add a delay ("archive N days after merge") instead of archiving immediately
- [FEATURE] Desktop sidebar: add a sort-direction toggle (ascending / oldest-first) to the session list
- [FEATURE] Claude Tag: wake on specific emoji reaction
- [FEATURE] Let a tool result declare itself single-turn, so it leaves the transcript after the model reads it
- [FEATURE] VS Code extension: filesystem path completion in @-mention (`../`, `~/`, absolute paths) like the CLI
- [FEATURE] Desktop sidebar: surface running tasks and unread conversations instead of collapsing everything into project groups
- [FEATURE] PermissionRequest hook input should include tool_use_id
- [FEATURE] Max (100x) plan is a must now!
- [FEATURE] Prompt for "Resume/Summary prompt" when approaching credit limit.
- [Feature Request] Add programmatic trigger mechanism for Claude Design sessions via API or webhooks
- [FEATURE] The /status command should include a timestamp
- [FEATURE] Make the markdown-table → Header:-value list fallback configurable
- Feature request: let Claude set the session color in response to a prompt (agent-invocable /color)
- [Feature Request] /insights date range filter (--since flag)
- [Feature Request] Support for older hardware/lower-end systems with newer models
- [FEATURE]
- Feature Request: Terminal profile dropdown in UI mode
- [FEATURE] Desktop app: setting to choose the integrated Terminal pane shell (e.g. Git Bash on Windows)
- in dont knopw which place to post this
- [FEATURE] Pull Enterprise Org Spend-Limit Data in Claude Code (CLI/statusLine)
- [Feature Request] Make agent cancellation safer with confirmation or pause mechanism
- [FEATURE] Dispatch / desktop-app mobile handoff is unavailable on Team/Enterprise plans — and fails silently
- [FEATURE] Cowork: allow enabling/disabling individual skills inside an installed plugin
- [BUG] TUI footer: status chip pushed onto its own line while the hint row is truncated
- [Feature Request] Exclude legitimate security work on owned code from security warnings
- Feature request: give PreToolUse hooks visibility into the in-flight assistant message (or flush it to the transcript before hooks run)
- [Feature Request] Clarify pricing value proposition for Claude Opus 5 given execution rate concerns FRAUDE FRAUDE FRAUDE LEGAL CASE
- Organizacao de trabalhos diversos
- [FEATURE] Keep conversations in the same order as before restart
- [FEATURE] Keep input while switching conversations
- [FEATURE] Detached session windows are missing the pane buttons (Views menu)
- Feature request: per-subagent skill allowlisting (not just preloading or full block)
- [Feature Request] Optimize CLI interface performance and reduce visual noise
- [FEATURE] Desktop: keyboard shortcut for Reply and Attach selection as context
- [FEATURE] Stop watermarking our code
- Voice dictation: allow typing while recording is active (don't stop the recording on keystrokes)
- [FEATURE] claude.ai Projects: let a session read across projects, and let projects be grouped
- [FEATURE] Claude Desktop (macOS): allow removing entries from the Recent projects list — `claude project purge` does not clear them
- [Feature Request] Emit the existing remote.session_id statusline payload field for local Remote Control sessions
- [FEATURE] VSCode Extension - Add support for /export command
- [Feature Request] Support for custom cryptographic primitives in code generation
- [FEATURE] Install "category" in marketplace
- Session picker should show continuation/chain linkage between transcripts
- Feature request: persistent "thread brief" artifact as a default mode for long strategic conversations
- [Feature Request] Integrate other LLM installed through CLI for multi-model routing in Claude Code
- [Feature Request] Add read-only API for conversation history and project documentation export
- [FEATURE] Add a "Prompt" / instructions field to cloud environment configuration
- [FEATURE] Threads: navigable in-session sub-conversations that inherit full context (fork + agent-teams UI + merge-back)
- Feature Request: German (Deutsch) language pack for the CLI UI
- [FEATURE] Keyboard shortcut to jump to the next tab awaiting input (yellow-dot / needs-attention)
- [Feature Request] Add option to customize or disable welcome banner in TUI
- Feature request: localize the built-in Plan Mode dialog when `language` setting is set
- [FEATURE] CLI command to list resolved skills without starting a session (no API call)
- [Feature Request] Improve multi-turn context reasoning and topic relationship tracking in Opus
- [FEATURE] Emit semantic marks for assistant-message boundaries and Stop-hook verdicts, so superseded drafts can be collapsed
- Built-in browser pane cannot reach private/internal origins — request an opt-in per-project allowlist and a clearer error
- [FEATURE] Reduce Auto mode stage-2 classifier latency with a bounded decision contract
- Filtering
- Feature request: per-model default effort levels
- [FEATURE] Bring the Code tab's side chat (/btw, Ctrl+;) to the Chat and Cowork tabs
- [FEATURE] Trusted devices: support multiple concurrently enrolled FIDO2 authenticators (currently single-enrollment, replace-only)
- [Feature] Allow workflow scripts to mark agent results as failed (red indicator)
- Feature request: setting to disable/hide the auto-generated prompt-suggestion list
- [Feature Request] Add built-in accessibility output style with answer-first formatting and complete sentences
- [Feature Request] Implement OODA loop pattern for agent decision-making
- [FEATURE] Desktop (Cowork): task panel auto-opens when the first chat message is sent; no setting to keep it closed
- [FEATURE] When running subagents, label each tool call with the subagent name (visual studio code)
- [FEATURE] Claude Desktop: make quitting discoverable instead of burying it in the hamburger menu
- [FEATURE] Make session transcripts/memory project-portable while keeping scratch files local-only (linked by session ID)
- [FEATURE] Bash mode variant that does not add the command and its output to context (e.g. `!!`)
- [FEATURE] Allow user to move a specific CC Routine run into the "sessions" CCW list
- [FEATURE] Configurable pageSize for AskUserQuestion tool
- [FEATURE] Live tokens-per-second (TPS) display during streaming
- [Feature Request] Multi-file multi-line code selection highlighting like Cursor IDE
- [FEATURE] Proposal: Treat Project Instructions as an Always-Active Skill
- [Feature Request] Setting to hide Edit/Write diff and content previews in the transcript
- [Feature Request] Add UI padding/density customization settings
- [FEATURE] Inline side-panel preview for .docx/.pptx/.xlsx sent via SendUserFile (pipeline already exists in Cowork)
- [FEATURE] Claude should accurately be able to respond to all claude billing/licensing request information regardless of the model
- [FEATURE] Allow individual users to opt out of org-level customized spinner verbs
- [FEATURE] Register Claude Desktop as a valid target for Windows 11's native search/Copilot key remap
- [FEATURE] Session pinning is invisible from the CLI: expose it in `agents --json` and `--help`
- effortLevel setting cannot persist the "Max" effort tier the UI offers