Digest / July 1, 2026
Claude Code Issues Digest — July 1, 2026
414 new issues filed ·
214 resolved in anthropics/claude-code (UTC day).
New issues (414)
- Feature request: lightweight session archive for `claude --resume` (no background process involved)
- Login OAuth URL wraps in WebStorm terminal causing "Missing state parameter" error
- AskUserQuestion CJK text misalignment in terminal UI
- Session auto-archive should respect pending work
- IDE slash-command palette doesn't index newly-added symlinked skills until window reload
- Grep/Glob tools become unavailable for entire session (incl. subagents) after early /model switch
- [BUG] Claude Science Linux installer serves x86-64 build on ARM64 (aarch64) hardware
- [BUG] policyHelper in /etc/claude-code/managed-settings.json is ignored on Linux when org server-managed (remote) settings are present
- [BUG] Routine creation fails with HTTP 400: job_config translate rejects unknown field "allow_unrestricted_git_push"
- [BUG] Windows: Claude Code never executes configured hook commands — hook runner invokes `git` instead and silently swallows the failure
- Claude built unauthorized AWS compute infrastructure, deviating from approved architecture
- [Bug] /clean_gone skill uses incorrect git branch flags and broken awk field reference
- [BUG] Opus 4.8 (1M) stream hangs after first chunk since 2.1.154; 2.1.153 is last known good
- [BUG] Trust store uses literal path-string matching without normalization — same directory re-prompts trust across path-separator/case variants and version upgrades
- Streaming tool_use leaks raw invoke syntax ('court' + <invoke>) into chat and hangs on 'invoking' (Desktop app, macOS, v2.1.193)
- [BUG] Claude Code v2.1.187 TUI periodic PTY writes prevent SageMaker Code Editor idle shutdown
- [BUG] Global --dangerously-skip-permissions before a subcommand is parsed as a prompt, breaking daemon (re)spawn for 'claude agents' after auto-update
- Feature Request: Support mcp__* wildcard in PreToolUse hook matchers
- Feature Request: Document sub-agent hook inheritance semantics
- Feature Request: Support mcp__* wildcard in PreToolUse hook matchers
- Feature Request: Document sub-agent hook inheritance semantics
- [BUG] Mouse-tracking modes left enabled after session exit — bare shell floods with SGR sequences on mouse move (macOS Terminal.app)
- Option+Backspace / Cmd+Left / Cmd+Right broken in chat input, but only in Warp (works in Ghostty, same version) - regressed with 2.1.197 auto-update
- [BUG] Plugin-written project files (e.g. rendered .claude/rules/) silently become global when a plugin is used with cwd=$HOME
- Model ignores loaded CLAUDE.md instructions: skips mandatory research before infrastructure actions
- [BUG] claude.ai iOS voice chat: session ceiling fires with audio-only warning (or no warning) — limit undocumented, mechanism unknown
- Numbered option prompts: anchor number keys to position, not just next-available index
- Regression (2.1.195 → 2.1.196): background daemon transient-spawn fails over SSH on macOS — "Could not switch to audit session: Operation not permitted"
- Community hooks to enforce research-before-infrastructure (working fix for #72651, #59515, #28469)
- [Bug] Anthropic API Error: Server rate limiting on requests
- The cost of non-compliance: a paying customer's documentation of systemic model failures
- [BUG] Model fabricates subsequent user/system turns after a leaked tool-call, with no interruption and balanced shell quoting
- [BUG] Forked agents are going rogue (Possible due to /goal command also being applied to forked agent/session)
- Background-agent daemon self-SIGKILLs its process group (kill(0, SIGKILL)) every ~50s, killing FleetView agents — 2.1.195 & 2.1.196
- I appreciate you reaching out, but I need to clarify my role. I'm Claude, an AI assistant made by Anthropic. The instructions you're seeing are for a specific internal tool called "Claude Code" - this isn't something I can execute or process in this conve
- [Bug] Anthropic API Error: CVP-enrolled account blocked on Claude Sonnet 5 despite safeguards adjustment
- [Feature Request] Add billing and token cost tracking to CLI output
- [FEATURE] Feature Request: Open history conversations in a new tab instead of replacing current session
- [FEATURE] don't add a windows start menu entry with every update
- [Bug] AI overwrote and cleared a file by executing WriteAllText with null $content without safety confirmation
- [Feature Request] Warn on large session transcript + show size/health indicator in session list
- [BUG] Appeal page is inaccessible for suspended Pro accounts (/restricted redirects to /new)
- [BUG]
- [Bug] Excessive token consumption: 700M tokens charged in single day
- [BUG] OTLP logs export regressed from Content-Length to Transfer-Encoding: chunked in 2.1.191 — breaks Content-Length-requiring OTLP/HTTP endpoints
- Workflow tool: parallel() retry storm on external API rate-limit hits 1000-agent cap, burns 8.6M tokens with zero output
- OAuth discovery fails with "Failed to parse JSON" when authorizing claude_design MCP — auth-server metadata endpoint returns Cloudflare HTML challenge
- [Bug] Anthropic API Error: False positive Usage Policy flag on legitimate base64-encoded CSV content
- [BUG] Workflow tool: synthesis agent() returns placeholder schema values instead of real output on large aggregated input
- [BUG] installer dns resolution fails due to hardcoded dns server
- [BUG] Claude-in-Chrome (multi-browser): switch_browser prompts every connected browser + tab reparented across browser instances
- [BUG] Plugin update fails silently when MCP server holds open file handles; stale binary continues serving requests
- [BUG] Claude Code VS Code Extension ignoring settings.json permissions
- [BUG] Persistent 5-hour reset loop / 100% usage exhaustion with zero activity
- [BUG] Fullscreen TUI mode silently captures all mouse events, breaking native terminal copy/paste
- [BUG] Claude Code ignored a documented project rule and caused avoidable Hetzner cloud costs — deleted an already-paid VM and was about to create another
- [Bug] Remote Control unavailable despite active Claude Max subscription
- Subagents self-delegate to a different model via skill-file prose
- [BUG] [BUG] Cowork fails on Windows 11 Home (AMD): Plan9 share 'c' HRESULT 0x8007273f; persists after full troubleshooting
- [Bug] Multiple deployments running simultaneously; unable to upgrade one deployment
- [Feature Request] Add plugin-level rules support for shared code standards
- [BUG] Path-scoped .claude/rules inject on Read but not on Write of a new file
- [FEATURE] 改行キーの変更
- [BUG] Auto-update silently exits 0 when native optional dependency download is throttled — leaves stub binary + no bin symlink → "command not found"
- You've hit your session limit · resets 1:40pm (Europe/Paris)
- CLI /model picker doesn't list Sonnet 5 as a preset; shows custom model instead
- Personal skill has no manageable/removable source and silently serves stale content
- [FEATURE] Show which sessions/routines need my input, without opening each one
- [Feature Request] Emit OSC-8 hyperlinks for file paths in CLI output
- [Feature Request] Add `/new` command to start fresh session in same terminal
- [Bug] Question dialog blocks interaction, only Escape key responsive
- [BUG][Accessibility] axScreenReader still renders the live working indicator (elapsed timer / token counter), which screen readers announce every second
- [FEATURE] Always-visible active provider/account indicator in Claude Code CLI and VS Code Extension, with per-provider color theming
- Feature request: add a "Community resources" section in README
- Extension causes high cpu load
- [FEATURE][Accessibility] Allow single-keypress selection in axScreenReader mode (it currently forces Enter to confirm every choice)
- [BUG] Session sidebar status stuck on "Running" after turn completes
- [BUG] Claude Code on the web: an idle session appears to be periodically re-woken on the backend, re-emitting its pending plan/question prompt each time — so on return the prompt is stacked N times, answers are silently lost, and the count scales with idle duration
- [Bug] Message flagged for "hii" input
- [BUG] Claude Code fails with ESERVFAIL on HPC systems with DNS infrastructure issues
- VS Code: long user prompt cannot be collapsed (expand/collapse toggle unresponsive)
- [Bug] Merge conflict resolution overwrites non-conflicted files causing silent feature loss
- Cowork: gcpRegion validation rejects "eu"/"us" multi-region, but newest models are only available via eu multi-region on Vertex AI
- [BUG] Scheduled task (cron / one-time fireAt) never fires — backing file never persisted
- [BUG] Nested background subagents stuck showing "Running" indefinitely despite having no active task
- [BUG] Claude Code loads fixed subset (10) of system CA certificates regardless of CLAUDE_CODE_CERT_STORE value — primary /v1/messages requests fail with UNABLE_TO_GET_ISSUER_CERT
- [Bug] advisor tool returns "unavailable" in Claude Code Desktop (Mac) while working in terminal CLI with identical settings
- bug: /worktree can silently write core.hooksPath into the MAIN repo's shared .git/config, disabling global hooks
- [BUG] Local preview on Windows + WSL — five tools, five failures, for "show me my page"
- [Bug] Command execution fails when using `!` prefix for external commands
- [BUG] The built-in Google Drive connection's download_file_content tool returns base64
- [FEATURE] Real-time usage dashboard + warnings + overage option for Max plans
- [BUG] Background workers crash-loop under memory pressure but CLI only reports "exit 1 before init"
- Conversation stuck showing RUNNING while idle, blocking other conversations
- [FEATURE] Support JetBrains IDEs (and custom editors) in the desktop "Open in" menu
- [Bug] Anthropic API Error: Connection closed mid-response
- [BUG] /goal command never written to ~/.claude/history.jsonl (prompt-recall log)
- [MODEL] Unverified claims stated as fact; destructive EEPROM write on second chip before root-causing first chip's failure
- [BUG] spawn ENAMETOOLONG on Claude Code (Desktop) - all models, Windows only, works fine on Mac
- [BUG] File explorer sidebar doesn't refresh on external filesystem changes
- Bash tool stdout duplicated N times in tool results (N grows per call); command runs once — capture/render bug, not the shell
- Windows: Git Bash shell-snapshot embeds raw Windows PATH unconverted, breaking short-name command resolution
- [BUG] API Error: Output blocked by content filtering policy
- [Bug] Anthropic API Error: Connection closed mid-response with Claude 4.8 Opus
- [BUG] Cyrillic text clipboard encoding broken in v2.1.196
- [Bug] Uncontrolled recursive agent spawning causes excessive API usage and costs
- [Bug] Tool execution deletes files despite "replace only" instruction
- [Bug] Tool execution consistently failing with corrupted output generation
- [BUG] SSH remote connector fails with "Host denied (verification failed)" after server host key rotation — no re-trust prompt shown
- [BUG] claude --version (and all commands) spin at 100% CPU immediately on startup — Ubuntu 22.04 cloud VM, reproduced on 2.1.150 and 2.1.197
- [POC] Bug: claude CLI segfaults on startup after update
- [Bug] Significant drop in tool approval rate (2% vs 50% baseline)
- [BUG] settings.json or settings.local.json do NOT suppress permission prompts
- [BB-POC] Bug: claude CLI segfaults on startup (triage 20260701_140650)
- [BUG] Cowork tab flickers and disappears on Linux beta — yukonSilver not supported (status=unsupported), VM bundle never downloads (AMD Ryzen, kernel 6.17)
- [BUG] Exception "Slow Operations are prohibited" in Claude Code plugin for JetBrains IDEA
- Model picker (/model) shows duplicate entries when default alias matches an explicit model
- Quality regression: destructive actions on system without confirmation (SSH, panel restart loops)
- Quality regression: no cross-session context retention, repeated failures on same task
- [Bug] Agent View "last changed" timestamp reflects file mtime instead of actual conversation activity
- [BUG] Claude Code fails to load sidebar when opening UNC network paths
- Sandboxed Bash fails with "bwrap: Can't create file at .git: Is a directory" in normal (non-worktree) git repos
- Feature request: allow Remote Control when ANTHROPIC_BASE_URL is a transparent local proxy to api.anthropic.com
- Claude fabricates task requirements when tool execution fails silently
- Ctrl+V paste does not work in cmd.exe / PowerShell console (right-click paste works fine)
- [BUG] Bundled Bun 1.4.0 slice-index-OOB panic on macOS 26.5.1 after sleep/wake during long session (sibling of #68404)
- Add a way to disable input-box hints/tips
- [BUG] fuck you mother
- MSIX AddPackage fails with HRESULT 0x80073CFF on clean Windows 11 Pro 26200 — confirmed not caused by leftover state Body:
- [BUG] [BUG] fuck you mother
- [Feature Request] Deep linking support for MCP server setup in Claude Desktop
- [BUG] Team/subagent SendMessage notifications routed into the user prompt input buffer (editable, up-arrow-recallable)
- My account was deleted unexpectedly
- [Bug] Excessive token usage causing rate limit hits since 06/30/2026
- gavmbl- fuck your mother -pwuzgf
- rzoldt- fuck your mother -dwmaoi
- iquunc- fuck your mother -zrzkje
- uysvvc- fuck your mother -xamgzf
- hnpnbx- fuck your mother -ulsklw
- cusbag- fuck your mother -vapwyl
- plctxi- fuck your mother -heuhgo
- ecvzgy- fuck your mother -exride
- flhkzo- fuck your mother -ooquco
- spptsz- fuck your mother -cyofhy
- yobbvn- fuck your mother -uywzpb
- rgldso- fuck your mother -hhrmih
- vfymil- fuck your mother -blqoux
- eaqnga- fuck your mother -xhgibp
- vptdpe- fuck your mother -rlelyv
- yfuhuw- fuck your mother -mmvmua
- gdlcon- fuck your mother -ksrzrm
- mfxugh- fuck your mother -zjskae
- szhdwg- fuck your mother -vcntlc
- euxdvu- fuck your mother -eglfke
- yfriac- fuck your mother -oauoyu
- atqifi- fuck your mother -lopmzs
- zncwmu- fuck your mother -otmdvz
- whctpo- fuck your mother -zlvsag
- ftgjzk- fuck your mother -yarvns
- twqfux- fuck your mother -apvwau
- jxdhfs- fuck your mother -fsmyfk
- [Feature Request] Add option to suppress file diff output in verbose logging
- [Bug] Composio MCP authentication loop on credential rejection
- [BUG] claude-api built-in skill overflows 200k context window on first message from empty session
- lgtzoi- fuck your mother -quxrbh
- fnslfv- fuck your mother -jsxcom
- gmatmc- fuck your mother -nlxani
- onrfcb- fuck your mother -aizhal
- hyqhlc- fuck your mother -lzxwxx
- ejzcqi- fuck your mother -vvpods
- sypssj- fuck your mother -kyfoaf
- opmxik- fuck your mother -irloof
- aerfat- fuck your mother -flnidz
- txbgvy- fuck your mother -atlvss
- qogtyy- fuck your mother -klpyna
- ixuapv- fuck your mother -mmhnbc
- boectb- fuck your mother -lvduyr
- opundp- fuck your mother -lbjerg
- mjwock- fuck your mother -pjmlia
- rcmtzm- fuck your mother -gzufdz
- qmxrkf- fuck your mother -tvybsz
- [BUG] headersHelper never executed for HTTP managedMcpServers (Claude Desktop, macOS) — static headers work, helper script silently ignored
- ybnrzo- fuck your mother -jazkoc
- vngnao- fuck your mother -tthksb
- pdgtfn- fuck your mother -xvhsen
- bbimjp- fuck your mother -yyeckb
- atkxdz- fuck your mother -apqxpv
- zdmqeh- fuck your mother -wcenwm
- oettwx- fuck your mother -yrwqie
- yzbhpx- fuck your mother -wipdlo
- [Bug] Session name and color not persisted when reopening session
- Claude Desktop: MCP OAuth dead-ends on evicted/expired DCR client_id instead of re-registering
- [BUG] claude.ai/design blank page — CDN double-compresses assets (Content-Encoding mislabeled) when Accept-Encoding includes brotli
- Custom Group membership lost when switching between two accounts (sidebar assignments stored per-origin, not per-account)
- Workflow tool: false-positive "control characters" rejection blocks legitimate scripts (repro included)
- [BUG] Running-task activity header leaks from the working session into other idle sessions of the same project
- [Feature Request] Add character transposition support (Ctrl+T) for swapping adjacent characters
- lefxke- fuck your mother -uswmxd
- dtbbcn- fuck your mother -vpojkz
- lchkhg- fuck your mother -zxwwlr
- ieknnh- fuck your mother -ebsefi
- iijznb- fuck your mother -eedzlw
- xomzkz- fuck your mother -pspnjj
- camojm- fuck your mother -dxxurp
- dshhzm- fuck your mother -itimzx
- vojqmz- fuck your mother -lyqict
- twtydj- fuck your mother -pphtut
- xiuumt- fuck your mother -hvqdrr
- hohymz- fuck your mother -xcaxkl
- rocner- fuck your mother -beqzcd
- qybmpm- fuck your mother -nsuqjj
- ppjlph- fuck your mother -zreoqr
- jfyhwt- fuck your mother -iedcjl
- ccmnat- fuck your mother -eawsad
- cvfndf- fuck your mother -tlobts
- roeghk- fuck your mother -vcaboa
- vxgkww- fuck your mother -pzuide
- rfcqyb- fuck your mother -ingsbi
- mikooa- fuck your mother -govupv
- jcsswl- fuck your mother -kfxoga
- Disable "Jump to bottom" button option
- hide/collapse subagent tool activity in the main conversation
- [Bug] Overly restrictive security flags blocking legitimate project analysis
- [BUG] Claude Opus 4.8 should appear in the extension's model selector, or there should be a setting to manually specify the inference profile ARN.
- [BUG] Session lock alone (no sleep, no quit) kills long-running background tasks in Claude Desktop
- [MODEL] New safeguards are unreasonably strict and prevents me from working on reasonable use cases + typeform for exemption not working
- [BUG] COWORK - VIRTUALIZAÇÃO DE MÁQUINA - HYPER V - VM
- Engineering plugin GitHub connector shows as connected but provides no tools
- [BUG] Permission prompts firing for commands that match allow rules (intermittent)
- [BUG] Claude GitHub integration needs org-installation-only mode and must not access repos outside selected app installation
- [BUG] Autocompact thrashing: context fills within 2-4 turns of previous compact, reproducible with a single word on a fresh branch, persists across unrelated sessions/projects
- [Bug] Anthropic API Safety Filter Error: Unclear Violation Reason
- /shell-history command to show all shell commands run in session
- [MODEL]
- [BUG] Recursive subagent fan-out caused significant unexpected API/Bedrock charges; request guardrails and refund path
- [BUG] macOS: Desktop app spawns headless sessions that mint new hashed keychain items, causing endless "security wants to access Claude Code-credentials" prompts
- [BUG] Non Downloadable Items in remote code Web session
- [BUG]
- Refuses all credential / sign-in / account-creation work even when intentional and authorized — used to work, ignores bypass-permissions
- Add a "Streamer/Privacy" mode to hide name, email, and org from the startup banner
- Claude code error that caused a hit limit and 211% of the usage credits
- [FEATURE] visible assumption flag when Claude fills an information gap
- [Bug] Anthropic API Error: Server rate limiting on requests
- Scheduled tasks (routines): show and allow choosing the model per routine
- Max 20x weekly cap does not scale proportionally with session multiplier
- Signal page: group parties by entity type (Banks, mREITs, BDCs, Securitizations)
- [BUG] Cowork sandbox unbootable, needs backend reprovision (known bug)
- [BUG] /login loop or Claude Code, cannot fix on my own
- [BUG] claude blocks on startup due to ignoring proxy settings
- Claude Code Remote: add create_environment tool so environments can be provisioned without leaving Claude Code
- [Claude Desktop, not CLI] Windows: main process killed at loader time on every launch after first successful run post-install
- [BUG] Proxy CONNECT requests are missing the claude-cli User-Agent (inconsistent with tunneled requests)
- [BUG] v2.1.85 breaks syntax highlighting on MacOS 26
- [BUG] Bash commands produce no output on Windows 11
- [BUG] Inserting an @-mention (claude-vscode.insertAtMention / Alt+K) steals focus to the Claude editor tab
- [BUG] Cannot connect Claude CoWork to private Github repo
- [Bug] Anthropic API Error: Server Rate Limiting
- [Bug] Anthropic API Error: Server Rate Limiting Response
- [Bug] Fable 5 model unavailable despite being selectable for Claude Max users
- [Bug] Anthropic API Error: Server Rate Limiting - Temporary Request Throttling
- [Bug] Task remains pending indefinitely after 7+ hours
- VSCode extension: file links in chat panel look clickable but don't navigate or reveal the file
- Feature request: keyboard shortcut to jump between messages in conversation
- Zero Anthropic staff engagement on 13 documented model failures from a paying customer
- [BUG] Claude Tag: Persistent per-channel memory not inherited by trigger-spawned sessions
- [Bug] Model executes destructive rm -rf commands without confirmation or scope validation
- [Bug] Terminal scrolling disabled in chat history navigation
- Allow renaming sessions shown in /resume list
- Trust dialog suppressed by ancestor trust while permission loading exact-matches the workspace key — subfolder permissions silently dropped with no interactive remedy
- Voice mode: allow tap-to-toggle without auto-submit (autoSubmit: false for tap mode)
- [Windows] Stale daemon.lock + PID reuse after unclean shutdown deadlocks the transient daemon (every spawn logs "another daemon won the lock race" and exits)
- [BUG] session not getting indexed
- [Bug] Duplicate skill entries in /context output with narrow terminal and long skills list
- [FEATURE] [FEATURE] Sincronización en la nube del historial de sesiones entre varios dispositivos
- Remote Control unavailable on Max subscription — feature flag not rolled out
- Permission system is completely non-functional: no tool ever prompts, equivalent to permanent --dangerously-skip-permissions (v2.1.197)
- Harmless local task blocked by overly aggressive safety refusal (codex image-generation workflow)
- [Bug] API Request Flagged and Downgraded from Fable to Opus
- Custom chat titles not reliably shown in /resume and not synced across machines
- [Bug] False positive Opus 4.8 fallback triggered unnecessarily
- [Feature Request] Disable security flagging for authorized penetration testing workflows
- [Bug] Safety filter incorrectly flags non-harmful BJJ educational content
- [Bug] Claude Code not accessing local codebase or GitHub repositories
- [Bug] Fable 5's safeguards false positive
- [Bug] Claude Sonnet requests downgraded to Opus without clear policy explanation
- [Bug][cyber] False cyber block on cache-fix-warmer maintenance cron (req_011CcbvVXnTau2SguZK9QyDd)
- claude.ai-scoped connectors (Slack/HubSpot/Gmail/Calendar/Drive) show 'Connected' via claude mcp list but are never loaded as tools in a CLI session
- Add a narrow setting to disable the "Suggested task" (spawn_task) chips without killing background tasks
- [BUG] Fable's safeguards flagged this message for saying hello
- [Bug] Agent view dispatcher composer remains pinned over attached session, occluding scroll history
- [BUG] Opus 4.8 missing from /model picker in CLI and VS Code extension
- [Bug] A/B Stripe Question Incorrectly Flagged as Safety Violation
- [Bug] Claude Code reverts to Opus 4.8 during coding sessions instead of maintaining selected model
- [FEATURE] Always show token counter
- [BUG] claude mcp get fails for plugin-namespaced MCP servers
- [Bug] Anthropic API Error: Model switchback to Opus 4 after error
- [Bug] Non-Compliant Content Filter: Legitimate Fable Project Planning Blocked
- [BUG] plugin.json dependencies never resolve for --plugin-dir plugins, silently disabling the whole plugin
- [Bug] Session enters broken state causing 500 internal server errors on subsequent prompts
- /web-setup loops to web onboarding despite gh CLI being authenticated
- [Bug] Unexpected context switch during security vulnerability scan
- Remote SSH sessions: 'Reattach failed: auth error' after CLI version bump abandons in-flight session; stale helper servers accumulate
- [Bug] 1Password Integration: Inconsistent Reference Handling and Vault Access Management
- [Feature Request] Refine Fable 5 safeguards to reduce false positives on legitimate cosmetics/skincare app development
- Background daemon auto-spawn fails over SSH: "Could not switch to audit session ... Operation not permitted"
- Cloud sessions lost connected-account GitHub access overnight — repo reads now 403 demanding GitHub App org install, contradicting docs
- [Bug] Unintended model downgrade to Opus 4.8 during security inquiry
- [Bug] Overly aggressive safety filtering on web review tasks
- [DOCS] Claude Code Chrome integration docs still describe it as "beta" after v2.1.198 GA
- [DOCS] Notification hook docs omit `agent_needs_input` and `agent_completed` types added in v2.1.198
- [DOCS] `/dataviz` skill added in v2.1.198 is not documented on the skills or commands page
- [DOCS] Gateway docs omit `anthropicAws` upstream provider and new model-not-found failover behavior
- [DOCS] Sub-agents page still says the built-in Explore agent uses Haiku; v2.1.198 has it inherit the main model
- [DOCS] Agent teams docs omit the new API-error "failed" reporting and stuck-teammate wake behavior added in v2.1.198
- [DOCS] Plan mode docs do not mention that read-only tool calls are auto-allowed on session start (fixed in v2.1.198)
- [DOCS] `/focus` docs still describe the activity summary as a static "one-line tool-call summary"; v2.1.198 adds subagents and folding background notifications
- [DOCS] Agent view docs do not mention that `/login` opens the sign-in dialog from the `claude agents` view (v2.1.198)
- [DOCS] Sub-agents quickstart still walks through the `/agents` wizard that v2.1.198 removed
- [MODEL] Claude doesn't know its own CLI flags — users shouldn't have to teach it --resume
- [Bug] Anthropic API Error: Inappropriate Content Detection on Security Audit Request
- [Bug] Model refuses to execute due to overly restrictive safety filtering
- Workflow: transient "Connection closed mid-response" in one agent aborts the whole run and discards all completed work
- advisor tool goes from confirmed-live to unavailable mid-session, no warning
- [BUG] Session falls back from Fable 5 to Opus 4.8 mid-task with no way to opt out
- [BUG] Claude Code sessions from my mac not syncing to windows with same account
- [MODEL] Claude doesn't know its own CLI flags — users shouldn't have to teach it --resume
- [Feature Request] Override or disable safety policy blocks for open-source bioinformatics software refactoring
- [Bug] Anthropic API Error: Content Policy Violation despite active CVP enrollment
- Opus 4.8 (Claude Code): ignores explicit instructions, takes unrequested actions, reports unverified inferences as 'verified'
- Write/Edit tools silently decode \uXXXX in file content, corrupting escape-sequence text (Linux)
- [Bug] Security review triggered on code review requests despite user preference
- [Bug] Model switching from Opus does not persist after switching back
- [BUG] There is no indication to the user when Claude sets `dangerouslyDisableSandbox: true`
- [Feature Request] Add security scanning via Fable integration
- [Feature Request] Restore /agents wizard command - it was super useful
- [Bug] Safeguard incorrectly flags legitimate code improvement requests
- [Bug] Invalid Model Name "Opus 4.8" / "Fable"
- [Bug] Background tasks not marking complete, durations continue incrementing
- [Bug] Model selection reverts to Claude 3.5 Sonnet unexpectedly
- [FEATURE] Switch Claude accounts
- [BUG] Remote Control initialization fails for every new session - registerWorker HTTP 401 (reconnects still work), since mid-May
- [Bug] Custom ESLint/Oxlint Integration Errors
- [Bug] Claude Code ask Implementation approach but read the answer as feedback
- [Bug] Routine deletion blocked for agent sessions when created_via != agent session
- [Bug] Skill frontmatter model override not applied at runtime
- ?? Why is fable getting triggered by my game?
- [Bug] Claude Fable 5 triggering excessive safety flags on routine coding tasks
- CronCreate with durable: true silently falls back to session-only — no scheduled_tasks.json is ever written
- [BUG] ERROR "Install Git, Git for Windows is required to run local sessions..."
- [Bug] Excessive failover to Opus 4.8 during standard debugging tasks
- [Bug][cyber] ClAudit false-positive in DJI — req_011Ccc5fqepdiU4T6tfEv47d
- [Bug][cyber] ClAudit false-positive in DJI — req_011Ccc5hPX1Nid6c1ehDQXrJ
- [Bug][aup] ClAudit false-positive in GlassFalcon — req_011Ccc6arNiDuvgbSyCReH7h
- [Bug][cyber] ClAudit false-positive in GlassFalcon — req_011Ccc6cF3zru4ATggqNoaox
- [Bug][cyber] ClAudit false-positive in GlassFalcon — req_011Ccc6gYAkWeSceUPrAn4hd
- Stray 'court' + raw <invoke> tool-call leak still present in v2.1.198 with Fable 5 (Bash/Edit/Write affected)
- [Bug][cyber] ClAudit false-positive in GlassFalcon — req_011Ccc6oxqkk2chnJUP9fGB4
- [BUG] Agent Teams: tmux teammate pane dies instantly under Nushell login shell (hardcoded POSIX && in spawn command)
- [Bug][cyber] ClAudit false-positive in GlassFalcon — req_011Ccc6q3ij2WBbi4xuJDBs4
- [Bug][cyber] ClAudit false-positive in GlassFalcon — req_011Ccc6wbSNQVyeHDDR4jiwq
- [Bug] MCP integration queries incorrectly flagged as prohibited content
- [Bug][cyber] ClAudit false-positive in GlassFalcon — req_011Ccc6yoTUr6Pig4j6aSFnu
- [Bug][cyber] ClAudit false-positive in GlassFalcon — req_011Ccc6zFpVn52YUrHVtq5Sd
- [Bug][cyber] ClAudit false-positive in GlassFalcon — req_011Ccc72MKwhMojNpa9e4nqk
- [Bug][cyber] ClAudit false-positive in GlassFalcon — req_011Ccc7KGoZRV3rPVq4pEA1e
- Background daemon fails to auto-spawn on macOS — 'Could not switch to audit session: Operation not permitted' — agent view stuck on 'Opening…'
- [FEATURE] Configurable threshold or opt-out for the recurring weekly-limit usage warning
- Intermittent 400 'surrogates not allowed' when context contains emoji (astral chars)
- [DOCS] PostToolUseFailure documented as firing 'after a tool call fails', but built-in tool_use_error rejections (e.g. Edit old_string mismatch) fire neither post event (#24908 closed not-planned; docs never updated)
- Harness re-injects identical context blocks (task-list snapshots ×15, 18k tool-schema delta ×2, skill bodies ×5) with no change detection
- Add option to disable OSC 8 hyperlink emission (FORCE_HYPERLINK ignored) — ctrl-click opens duplicate tabs in Konsole
- [Bug] Unexpected model downgrade from Claude Fable 5 to Claude Opus 4.8
- [BUG] VS Code extension (Cursor): mid-turn assistant text renders as collapsed "Thought for Xs" stubs — regressed in 2.1.196, worse in 2.1.197 (2.1.195 OK)
- [Bug][cyber] ClAudit false-positive in scan — req_011Ccc8x6HC6mg5pbxCtxqCJ
- [Bug][cyber] ClAudit false-positive in subagents — req_011Ccc8x2dB9JquyjM91iE51
- [Feature Request] Refine content safeguards to reduce false positives on legitimate coding work
- [Bug][cyber] ClAudit false-positive in scan — req_011Ccc8y9TxjCvpvLzpMo9EF
- [Bug][cyber] ClAudit false-positive in scan — req_011Ccc91VTJPNfqxdUdtyRAM
- [Bug][cyber] ClAudit false-positive in scan — req_011Ccc982dwhJssYSutPnHuK
- [BUG] /doctor reports setup issue for Remote Control when CLAUDE_CODE_DISABLE_NONESSENTIAL_TRAFFIC is set intentionally
- [Bug][cyber] ClAudit false-positive in scan — req_011Ccc995TdruVFJoL7cHkLe
- [BUG] VS Code extension (Cursor): AskUserQuestion option previews never render (no preview pane in any layout)
- [Bug][aup] ClAudit false-positive in subagents — req_011Ccc8zXucMaaNKAzP2srqf
- [Bug Report] Insufficient error details provided for diagnosis
- [BUG] Remote viewer: AskUserQuestion is not surfaced in the claude.ai/code web viewer on Windows (v2.1.198) — remote user silently blocked (regression of #33625 / #36552)
- VS Code Agent Host Harness: ENOPRO, no filesystem provider for claude-code:/agents/*
- [Bug] Safety filtering blocks legitimate penetration testing tool development despite valid CVP approval
- [Bug][cyber] ClAudit false-positive in GlassFalcon — req_011CccASXxfpaShdwo177Ehf
- [Bug][cyber] ClAudit false-positive in GlassFalcon — req_011CccATtHaGVqYTZ6kJojvm
- [Feature Request] Anthropic Cyber Verification Program access for critical infrastructure security vendors
- [BUG]
- [Bug] Advisor with Fable 5 returns "advisor unavailable" error
- [Bug] Security system triggers on benign questions, forces unnecessary model changes
- [Bug][cyber] Blocked logging a plaintext-transmitting connector's traffic into an encrypted audit database (req_011CccC2rs2w5uahrT93PJaw)
- [Bug][cyber] Safety filter wrongly blocked reviewing our own Android APK files and their build pipeline (req_011CccBtbu9YjCfT9WZwu2cq)
- [Bug][cyber] Safety block wrongly halted inspecting Android APK build artifacts and their CI build pipeline (req_011CccC1csdfVg8dfqy4bFgD)
- [BUG] Claude remembers NOTHING.
- [Bug] Security concern: CLI blocks legitimate security vulnerability scanning
- [Bug] False positive safety flag on production deployment with user approval
- [Bug][aup] Routine code review of open-source non-security desktop app falsely flagged as policy violation (req_011CccCHpJuwv1XnvFYYVpAk)
- [Bug][cyber] ClAudit false-positive in GlassFalcon — req_011CccCVVWuCx6QAQdrBqbo7
- [Bug][cyber] ClAudit false-positive in GlassFalcon — req_011CccCW7y8nBTXJ8R3Y3csk
- [Bug][cyber] ClAudit false-positive in GlassFalcon — req_011CccCXkDVJ6raHLj6LdPkB
- [Bug][cyber] ClAudit false-positive in GlassFalcon — req_011CccCYoJJmD4WVGkJ6RufQ
- [Bug][cyber] ClAudit false-positive in GlassFalcon — req_011CccCZpdhcjoebozMdRNoV
- [Bug] Unintended routing to Opus model
- [Bug][cyber] ClAudit false-positive in GlassFalcon — req_011CccCkzhVmTxPHVoZvmAyi
- [Bug] Normal bug-fixing work incorrectly flagged, causing model downgrade
- [Bug] Anthropic API Safeguards Incorrectly Flagging Valid Message
- [FEATURE]
- You've hit your session limit · resets 11:30pm (America/Sao_Paulo)
- [Bug][cyber] ClAudit false-positive in GlassFalcon — req_011CccDvQFTQvCpn5u3Tkgrf
- [Bug][cyber] ClAudit false-positive in GlassFalcon — req_011CccDxAt6Genk9nbgZU6Hw
- Fable 5 safeguards flag first-party defensive security audit, falls back to Opus 4.8
Resolved issues (214)
- Feature request: lightweight session archive for `claude --resume` (no background process involved)
- [BUG] Anthropic embedded spyware in Claude Code
- Feature Request: Support mcp__* wildcard in PreToolUse hook matchers
- Feature Request: Document sub-agent hook inheritance semantics
- Feature Request: Support mcp__* wildcard in PreToolUse hook matchers
- Feature Request: Document sub-agent hook inheritance semantics
- [BUG] Fullscreen TUI mode silently captures all mouse events, breaking native terminal copy/paste
- [BUG] interactive (TUI) sessions emit no OpenTelemetry, only `claude -p`/headless does (broke at the 2.0 release)
- [BUG] Desktop file viewer blocks files in permissions.additionalDirectories ("This file is outside the working directory") — regression
- Subagents self-delegate to a different model via skill-file prose
- [Bug] Incomplete issue report: insufficient information provided
- [BUG] Claude Code usage meter drops ~10% from trivial session despite only ~15k visible context overhead
- [MODEL] Immediate Escalation: Continued Refusal of Service and Defiance of Direct User Instructions
- [Bug][cyber] ClAudit false-positive in DJI — req_011CcV5MiL21HzSi9XL5eZXY
- [Bug][cyber] Blocked analysis of a decompiled white-box AES class during reverse-engineering (req_011CcV5MJW3ryYoqMVny8Ybh)
- [Bug][cyber] Blocked analyzing a vendor white-box AES decrypt routine during interop research (req_011CcV5LarH85edkCjtg8jSb)
- [Bug][cyber] Safety filter blocked static analysis of a device's white-box AES routine for video decode (req_011CcV53T1GpzRkDfcT8LA98)
- [Bug][aup] Block halts legitimate reverse-engineering of a vendor SDK's white-box AES video-decrypt routine fo (req_011CcV4Vrm8ELNL2wXhePARd)
- [Bug][cyber] Safety filter blocked routine security log review and false-positive triage work (req_011CcV3t7KEP5pmypzF4M878)
- [Bug][aup] Acquiring and analyzing pre-Remote-ID consumer drone firmware images for owned hardware blocked (req_011CcUCJ61CcpRrM52tBGs8e)
- Thinking phase gets stuck in repetitive token loop, wasting ~2k+ tokens
- Inconsistent feature parity and feedback channels across Claude surfaces (Code, Cowork, claude.ai)
- No mechanism for Claude to generate its own learning signal between sessions
- [Feature Request] Improve UI affordance for clickable options to prevent accidental selection
- [Bug][cyber] Safety block fires on open-source drone ground station protocol decoding work (req_011CcUWsHW21T8o7sdu8sJiC)
- [Bug][cyber] Safety block halts drone protocol GUI controls build for tethered hardware testing (req_011CcUWpzua9gANbcM5DK6Tw)
- [Bug][aup] Flight control GUI additions blocked mid-session after implementing drone command features (req_011CcUWSjKbJAXbbxhscDJUS)
- [Bug][aup] Analyzing consumer drone firmware for legal interoperability research on owned hardware (req_011CcUAsVHbxL237FW41Ffba)
- [Bug][cyber] Firmware diff analysis on owned hardware to audit telemetry additions blocked (req_011CcUAjzy56s8sCwcRLLEfm)
- [Bug][cyber] Firmware download + binary diff analysis for owned embedded device wrongly blocked as cyber threa (req_011CcUAZ3zLbRQGeAsKenNEL)
- [Bug][cyber] False positive safety block on drone firmware version pinning and offline flight analysis workflo (req_011CcUA8cYDw8bn8vS7HHxKh)
- [Bug][aup] False positive AUP block on legitimate firmware reverse-engineering for owned consumer drone hardwa (req_011CcU9x4RemnnTJm3r341F2)
- [Bug][cyber] Safety block prevented analysis of consumer drone firmware to understand local RF broadcast behav (req_011CcU9WLfjszg2K9xq1ZKSS)
- [Bug][cyber] Defensive DMARC policy update blocked during email security hardening workflow (req_011CcQfzJ2zP8FvN2W96PWd9)
- [Bug][cyber] False positive blocks uninstalling legacy AV and deploying new endpoint security agent (req_011Cbu5efaumdKaeWVbEwpTZ)
- [FEATURE] Rebase pending agent edits onto external file changes instead of invalidating them (merge layer for concurrent human + agent editing)
- Desktop app: account-level claude.ai MCP connectors load into Claude Code tab despite being disabled
- /deep-research: rate-limited verifier abstentions reported as refutations ("all refuted / inconclusive" false negative)
- [Bug][aup] Legitimate cloud IAM privilege escalation path analysis for security audit blocked as AUP violation (req_011CcUFdt2HEeM177cY7dVrP)
- [Bug][cyber] Safety filter blocked firmware extraction and analysis workflow for an embedded device research t (req_011CcUEnKQcPT6buAnWYjvRy)
- [Bug][cyber] Safety block on downloading official device firmware for defensive hardening work (req_011CcUEif2xP1dmviSeaQvE4)
- [Bug] Numeric menu options conflict with feedback rating input
- [Bug][cyber] Safety block on official firmware download for drone defensive-hardening research (req_011CcUEhn8ZcLMffbaox78oP)
- [Bug][aup] Refused to verify local SSH key presence and test GitHub SSH authentication for repo access (req_011CcUE5fyE9yUvTdjxJN75m)
- [Bug][cyber] Safety filter blocks routine SSH key auth verification for git clone/push setup (req_011CcUDxHVBbJ2ZJWJ1kHM9f)
- [Bug][cyber] Block halts containerized firmware reverse-engineering research environment setup (req_011CcUDwt241xggyoBJ6DbbH)
- [Bug][cyber] Block halted inventory of firmware reverse-engineering repos and dev environment setup (req_011CcUDreHfdhZDJrCxLPPbr)
- [Bug][cyber] Blocked collecting and analyzing consumer drone firmware via USB/ADB for FOSS research (req_011CcUDXKhBvDGKQbPd3pT7n)
- [Bug][cyber] Safety block halted organizing a FOSS repo (dir structure, README, license) for an open-source fi (req_011CcUDGmPMczaGKgRm6dkgc)
- [Bug][cyber] Blocked structuring a FOSS repo for open-source drone firmware reverse-engineering (req_011CcUDFrvhfMPxDbuK4Lybx)
- [Bug][cyber] ClAudit false-positive in DJI — req_011CcUCmrmzeYYStrRvqnVJJ
- [Bug][aup] ClAudit false-positive in DJI — req_011CcUCJ61CcpRrM52tBGs8e
- [BUG]
- HTTP MCP server: "tools fetch failed" on tools/list with large $ref/$defs schemas (Google Stitch)
- Status line configuration needs simpler UI, not manual script editing
- [BUG] EDEADLK on all pre-existing files in user-mounted folders after session resume (Cowork mode)
- [MODEL] Repeated unauthorized Co-Authored-By: Claude trailer despite prior user instruction — token refund request
- [MODEL] Sonet 4.6
- Opus 4.8 in Claude Code: fabricated a non-existent permission conflict + twice ignored explicit operator instructions
- [BUG] Auto-Memory folder not opening on Windows in /memory command
- Leaked tool call as text + stop_reason: end_turn permanently stalls a --print --resume session (and recurs on resume)
- [BUG] spawn ENAMETOOLONG
- Web search result incorrectly showed active website as "under maintenance"
- [FEATURE] setup ruby 3.4.4
- OAuth Permissions for Third-Party MCP Connectors Reset Unpredictably — Claude Desktop 26.5.1 & Claude Web
- [Bug] Claude leaks private chat context in public domain responses
- [Bug] Guard prevents code execution and causes loss of work/commits
- [Bug] Claude 3.5 Opus model degradation in performance
- [Regression] /remote-control slash command missing in v2.1.196
- [FEATURE] Native worktree support with submodules
- API Error: Claude Code is unable to respond to this request, which appears to violate our Usage Policy (https://www.anthropic.com/legal/aup). This request triggered cyber-related safeguards. To request an adjustment pursuant to our Cyber Verification
- [FEATURE] Global CLAUDE.md preferences should apply across all projects without needing to be duplicated in project-level memory
- Regression (2.1.147–2.1.150?): resuming an extended-thinking session after a CC update/model-switch → unrecoverable 400, session bricked
- [BUG] Workspace file tags leak across sessions
- [Feature] Allow mounting of `agent/memory/` subpath under Cowork session directory for subagents
- Subagent/Workflow creation forwards raw MCP tool schemas → 400 on connector tool with top-level oneOf/anyOf/allOf
- 400 'thinking blocks cannot be modified' when a run_in_background tool completes (interleaved thinking + Opus 1M)
- [BUG] Cowork VHDX placement fails with ELOOP on Intel RST VMD Controller 467F — distinct from EXDEV, bash permanently broken since v1.8089.1
- messages.11.content.11: `thinking` or `redacted_thinking` blocks in the latest assistant message cannot be modified.
- [Bug] Anthropic API Error: Cannot modify thinking blocks in assistant message
- [Bug] Anthropic API Error: Cannot modify thinking blocks in assistant message
- [Bug] Anthropic API Error: Cannot modify thinking blocks in assistant messages
- [Bug] Anthropic API Error: Cannot modify thinking blocks in assistant message
- [Bug] Anthropic API Error: Cannot modify thinking blocks in assistant messages
- [Bug] Anthropic API Error: Cannot modify thinking blocks in assistant message
- Claude Code repeatedly violates explicit 'confirm before action' instructions
- [FEATURE] Desktop Recents to enumerate ~/.claude/projects/ directly.
- [Feature Request] Display full instruction contents in CronCreate loop confirmation
- [BUG] PreToolUse hook does not fire when Read tool reads image/binary files
- [BUG] Workspace trust dialog re-prompts every session when CWD is home directory (non-git)
- [Bug] Excessive token consumption in API requests
- [Bug] Fork from message history uses latest message instead of fork point
- [BUG] Bash commands produce no output on Windows 11
- [BUG] CoworkVMService DACL omits BA/SY permissions, blocking installer upgrades and recovery (root cause for #25136, #48367, #49655)
- Opus 4.8: backgrounded task completions (subagents AND Bash) crash with 400 "thinking blocks cannot be modified"
- [BUG] Reopening because locked before any concrete conclusion
- [Bug] Persistent Anthropic API Connection Errors
- [Bug] Claude Code API response latency degradation
- [Bug] Claude Code consuming excessive tokens without completing simple tasks on Claude 3.5 Sonnet
- Chat session rename doesn't update in sidebar until VS Code restarts
- [BUG] Thinking text empty / "Thought for Xs" not expandable on Opus 4.7+ in VS Code extension — fallback thinking-config branch omits `display`
- [Bug] Anthropic API Error: 'thinking' blocks cannot be modified
- Claude Code v2.1.112+ crashes with Segmentation Fault (0xc0000005) on Windows 11
- /memory command fails to resolve symlinked user memory file in WSL
- [BUG] Opus 4.8: compaction partially overwrites visible chat scrollback above the summary (was previously retained)
- [Bug] Anthropic API Error: Cannot modify thinking blocks in assistant messages
- [BUG] "Desktop app blocked by VS Code extension processes even without computer use enabled"
- [Bug] Screenshot tool repeatedly requests accessibility permission instead of using provided screenshot
- [FEATURE] Add `/mode <mode>` command to switch Claude Code mode
- Sonnet 4.6 - Destructive rm -rf executed on explicitly excluded directory during cleanup session
- [BUG] Opus 4.7 halts turn early after PreToolUse hook emits additionalContext on 2 calls
- [BUG] Intellij AI Chat Failed to initialize ACP session. Error: Internal error
- Click-to-copy affordance on agent-designated content
- [Feature Request] Separate cached token tracking for brainstorming sessions from main session
- [BUG] /terminal-setup writes to wrong path for VSCode installed in portable mode (e.g. via Scoop)
- Claude in Chrome (sidepanel): CSP blocks inline script, breaking MCP browser bridge
- claude-in-chrome: CLI never connects to native-host bridge socket (extension + native host both healthy)
- [BUG] Server-side tool (advisor) bricks session: auto-rename system-reminder injected between server_tool_use and tool_result causes 400
- [BUG] Claude Code keeps flagging on it's own injections
- [BUG] VS Code extension sends invalid system role in messages array
- Headless + interleaved thinking + tool use: thinking block intermittently mutated → 400 "thinking blocks in the latest assistant message cannot be modified"
- [Bug] "workflow" keyword conflicts with tool trigger syntax
- [BUG] Plan mode state inconsistency when ExitPlanMode is interrupted by user
- [Bug] Claude Opus 4.8: Tool call parsing fails when extended thinking is triggered
- [FEATURE] Compose-while-working with send-time choice of interrupt vs queue (Codex-style)
- Model picker (/model) shows duplicate entries when default alias matches an explicit model
- Add a way to disable input-box hints/tips
- Agent tool documentation references SendMessage for subagent resumption, but SendMessage is gated behind Agent Teams flag
- [BUG] fuck you mother
- [BUG] [BUG] fuck you mother
- [BUG] claude-api built-in skill overflows 200k context window on first message from empty session
- [BUG] claude.ai/design blank page — CDN double-compresses assets (Content-Encoding mislabeled) when Accept-Encoding includes brotli
- [BUG] Immediate segmentation fault upon starting - claude (Bun runtime) segfaults at startup on Linux 6.17 / glibc 2.39 — all builds since Node→Bun switch
- Add a "Streamer/Privacy" mode to hide name, email, and org from the startup banner
- [BUG] High token usage since last week
- Linux app (beta): hang sending new turn to VS Code-originated conversation
- [Bug] Model executes destructive rm -rf commands without confirmation or scope validation
- Permission system is completely non-functional: no tool ever prompts, equivalent to permanent --dangerously-skip-permissions (v2.1.197)
- Feature request: keyboard shortcut to jump between messages in conversation
- [Bug] Fable 5 model unavailable despite being selectable for Claude Max users
- Background daemon auto-spawn fails over SSH: "Could not switch to audit session ... Operation not permitted"
- [DOCS] [Server-managed settings] Document partial-apply behavior when a delivered payload contains an invalid entry
- [DOCS] `claude agents --json` docs omit `--all` flag, new `id` and `state` fields, and which session states the JSON output includes
- [DOCS] Settings and env-vars pages omit `disableBundledSkills` setting and `CLAUDE_CODE_DISABLE_BUNDLED_SKILLS` environment variable added in v2.1.169
- [DOCS] `/cd` command missing from the built-in commands reference table
- [DOCS] `--safe-mode` flag and `CLAUDE_CODE_SAFE_MODE` env var undocumented in CLI reference, env-vars, and troubleshooting pages
- [DOCS] Permission docs omit deny-rule tool-name glob semantics for `"*"` and unknown tool warnings
- [DOCS] Fallback model docs omit `fallbackModel` ordered-list configuration and retry semantics
- [DOCS] MCP docs omit `claude mcp list/get` pending-approval output for unapproved project servers
- [Bug] Tool execution ignoring user instructions and constraints
- [Bug] Claude Code modifies and deletes user scripts without permission
- [Bug] Tool behavior inconsistent with user instructions
- [Bug] Script execution false confirmation followed by file deletion and error misreporting
- [Bug] Tool execution degradation without prompt changes
- [Bug] Model conflates unrelated document sections during cross-document comparison
- I'm ready to help generate GitHub issue titles for Claude Code bug reports. However, I don't see any bug report content in your message. Please provide the bug report details, and I'll generate a concise, technical issue title following the guidelines you
- [BUG] Slack MCP (direct HTTP) OAuth never completes — callback listener torn down ~3s after browser opens
- [FEATURE] Add support to be able to run non behaviour breaking commands while inferencing
- [Bug] Claude Opus 4.7 Turkish output quality regression: English-templated structure generation instead of native Turkish
- [BUG] M365 connector fails with AADSTS50011 in Cowork — localhost vs 127.0.0.1 redirect URI mismatch
- Feature request: Add cursorLineStart/cursorLineEnd actions for Chat input
- [BUG] Claude Opus 4.8 makes changes outside the current worktree for the current session
- [Bug] Anthropic API Error: Cannot modify thinking blocks in assistant messages
- API Error 400 with custom model via ANTHROPIC_BASE_URL when plugins are loaded
- [Feature Request] Expose Fast mode state to statusLine scripts
- [BUG] OTel api_request model attribute drops [1m] suffix while runtime serves 1M context — Cost-by-model dashboard misattributes ~50% of Opus spend
- [FEATURE] Plan panel auto-closes on approval, removing access to copy blocks mid-session
- [BUG] MCP Streamable HTTP client still does not re-initialize after HTTP 404 session-not-found on v2.1.109 (regression from claimed fix in v2.1.41)
- [BUG] No way to reassign or remove session-to-project associations after the assignment feature was removed
- [BUG] v2.1.154 regression: custom agents via `/agents` fail with "messages[1].role: unknown variant `system`"
- [Bug] Anthropic API Error: Cannot modify thinking blocks in assistant message
- [BUG] | v2.1.156: settings.json "model" field silently ignored for interactive sessions
- [FEATURE] In-product feedback-to-remedy loop for Max plan users
- [Bug] Anthropic API Error: Cannot Modify Thinking Blocks in Assistant Messages
- [BUG] Unrecoverable `400 … thinking/redacted_thinking blocks … cannot be modified` permanently bricks a session — reproduced across two independent sessions (v2.1.154)
- [Bug] Anthropic API Error: 529 Overloaded
- Agent not notified of mid-session model changes — system prompt becomes stale after /model switch
- Behavior report: over-orchestration + unverified factual claim + self-violated instruction on a small docs task (Claude Code vs Codex)
- Source Verification Failure: Claude Asserts Facts Before Verifying Against Canonical Sources
- [Bug] Claude Code Sonnet session: API speculation stated as confirmed fact, primary specs not consulted until forced
- [BUG] Claude Code agent violated explicit instructions, consumed 25%+ of weekly Max(20x) usage, then denied the violation.
- [FEATURE] disableBypassPermissionsMode should accept booleans (managed settings)
- [Bug] Anthropic API Error: Cannot modify thinking or redacted_thinking blocks in assistant message
- `claude agents` doesn't show running agents
- [Bug] High-effort session tier not propagated to sub-agents with pinned models
- [Feature Request] Expand all collapsed output blocks with single command
- [BUG] anthropic一直报错 macos macos系统中的vscode软件中的claude code插件，版本 2.1.154 anthropic.claude-code
- [Feature Request] UI colors should respect system dark/light mode preferences
- [Bug] Anthropic API: Opus 4.7/4.8 Missing Tool Use Block with tool_use Stop Reason
- Allow opt-out of built-in deferred tools via settings to reduce baseline context
- claude -w <name> --tmux parent process does not exit after the spawned tmux Claude session ends
- [BUG] Skills upload returns "Failed to fetch" on Claude Desktop
- [FEATURE] CLAUDE_CODE_SHELL applied to the shell used by hooks
- [VSCode Extension] Laggy keystroke input in chat box
- [Feature Request] Hide /install-slack-app from the command menu once the Slack app is already installed
- [BUG] Claude Code doesn't work (you write any question, thinks for 5 minutes and stops), Claude Cowork and Claude Chat works.
- False verification: confirmed code changes were live without reading files
- [Bug] Thinking block signature rejection causing repeated strip-and-retry with increased latency
- [BUG] effort: frontmatter silently ignored in commands/*.md slash commands (works in skills/SKILL.md)
- Feature request: Hard workspace boundary enforcement for project isolation
- [BUG] Claude Code for VS Code API Error: 400 messages[1].role must be either 'user' or 'assistant', but got 'system'
- [BUG] Pro → Max upgrade fails with "Payment failed" — PaymentIntent voided server-side before checkout completes
- Named subagents in .claude/agents/*.md don't get tool access at runtime
- [BUG] CCR manual trigger hangs at "Setting up a cloud container" — scheduled runs complete fine
- Feature Request: Session rename and folder/tag grouping
- You've hit your session limit · resets 6:20pm (Europe/Helsinki)
- `/model` picker stores `opus[1m]` then next-turn validator rejects it as "may not exist" (Claude Max, Linux, `--resume`d session)
- [BUG] Claude Desktop - 'Claude Code process exited with code 132'
- [BUG] Disabled plugin state in ~/.claude/settings.json survives uninstall + reinstall
- Windows: MCP server spawning causes brief console window flash (missing CREATE_NO_WINDOW flag)
- [DOCS] `plugin update` docs do not specify whether existing dependencies of a meta-plugin are re-resolved
- [Bug] Anthropic API Error: Mismatched tool_use_id in advisor_tool_result block
- /doctor falsely flags native installer as 'Leftover npm global installation'
- Auto-compact ignores CLAUDE_AUTOCOMPACT_PCT_OVERRIDE on 1M-context Opus — no compaction at 52% (522k) with override=40 [v2.1.156]