Claude Code

[BUG] naming convention should not use "-" as the replacement for "/" as it creates file and directories that start with "-".

### Preflight Checklist - [x] I have searched [existing issues](https://github.com/anthropics/claude-code/issues?q=is%3Aissue%20state%3Aopen%20label%3Abug) and this hasn't been reported yet - [x] This is a single bug report (please file separate reports for different bugs) - [x] I am using the latest version of Claude Code ### What's Wrong? Bash(rm /nhome/aw

[FEATURE] Better Multi Agent Management Visualization

### Preflight Checklist - [x] I have searched [existing requests](https://github.com/anthropics/claude-code/issues?q=is%3Aissue%20label%3Aenhancement) and this feature hasn't been requested yet - [x] This is a single feature request (not multiple features) ### Problem Statement For the Claude IDE - color code agents. more agent organization. how many times h

effortLevel in settings.json not honored on session start

## Summary `effortLevel` set in `~/.claude/settings.json` (user settings) does not appear to be honored when a new session starts. The active effort level seems to vary between sessions even though the persisted `effortLevel` value is unchanged by the user. ## Environment - Claude Code via VS Code extension - Model: Opus 4.8 (1M context) - Platform: Windows

[FEATURE] "How is Claude doing this session" feels meaningless

### Preflight Checklist - [x] I have searched [existing requests](https://github.com/anthropics/claude-code/issues?q=is%3Aissue%20label%3Aenhancement) and this feature hasn't been requested yet - [x] This is a single feature request (not multiple features) ### Problem Statement I often get asked "How is Claude doing this session?" - this seems to be a meanin

[Bug] Console billing: Tax calculation not resolving in checkout form

**Bug Description** Subject: Bug Report: Credits Purchase Checkout Stuck — Tax Calculation Not Resolving (Safari) Hi Anthropic Support, I am unable to complete the credits purchase on console.anthropic.com. The checkout form is fully filled in but the 'Buy $5 of credits' button remains inactive because the tax calculation never resolves — it stays at '$--' i

[BUG] Claude Code (VS Code ▎ extension) — copying a code block ▎ injects newlines into long commands

### Preflight Checklist - [x] I have searched [existing issues](https://github.com/anthropics/claude-code/issues?q=is%3Aissue%20state%3Aopen%20label%3Abug) and this hasn't been reported yet - [x] This is a single bug report (please file separate reports for different bugs) - [x] I am using the latest version of Claude Code ### What's Wrong? In the Claude Cod

channels: mid-turn inbound injected into TUI input box instead of queuing as discrete turn

## Summary When a message arrives on a `--channels` channel (e.g. Telegram) **while the Claude session is mid-turn**, the channel plugin injects the text directly into the Claude Code TUI **input box** (`❯ <text>`) rather than queueing it as its own discrete turn. ## Reproduction 1. Start a session with `claude --channels telegram` (or any channel plugin). 2

Feature request: copy button for individual chat messages

## Problem Codex and other AI chat interfaces in VS Code have a dedicated copy button that appears on hover for each message. Claude Code extension lacks this — the only way to copy a full response is manual text selection, which is tedious especially for long outputs. Code blocks already have a copy icon. The same pattern should apply to prose responses. ##

claude.yml workflow triggers on substring match, fires on incidental @​claude* strings in quoted text

## `claude.yml` workflow triggers on substring match, fires on incidental `@claude*` strings in quoted text ### Summary The `claude.yml` workflow's trigger condition uses `contains(github.event.comment.body, '@claude')`, which is a **case-insensitive substring match**. This fires on any comment containing the character sequence `@claude` anywhere in the body

...

Windows: path-scoped Write/Edit permission rules never match — Write tool absolutizes file_path pre-check and absolute Windows paths match no documented pattern form

### Preflight Checklist - [x] I have searched [existing issues](https://github.com/anthropics/claude-code/issues?q=is%3Aissue%20state%3Aopen%20label%3Abug) and this hasn't been reported yet - [x] This is a single bug report (please file separate reports for different bugs) - [x] I am using the latest version of Claude Code ### What's Wrong? ## Environment -

[BUG] The GitHub Actions integration on this repo closes threads as "stale" without warning

### Preflight Checklist - [x] I have searched [existing issues](https://github.com/anthropics/claude-code/issues?q=is%3Aissue%20state%3Aopen%20label%3Abug) and this hasn't been reported yet - [x] This is a single bug report (please file separate reports for different bugs) - [x] I am using the latest version of Claude Code ### What's Wrong? The github action

CLAUDE.md rules aren't reliably followed; PreToolUse hooks are a partial, hand-rolled stopgap

## Summary `CLAUDE.md` / `AGENTS.md` instructions are loaded every session but treated as context, not enforced — and in practice the agent violates even its own MUST-rules within the same session that authored them. `PreToolUse` hooks are the documented escape hatch, and they work — but only for rules that are *mechanically detectable from a tool's input*,

[BUG] Mode is not sync on remote-control session between CLI in server and in Claude Desktop or Claude Mobile

### Preflight Checklist - [x] I have searched [existing issues](https://github.com/anthropics/claude-code/issues?q=is%3Aissue%20state%3Aopen%20label%3Abug) and this hasn't been reported yet - [x] This is a single bug report (please file separate reports for different bugs) - [x] I am using the latest version of Claude Code ### What's Wrong? Claude code mode

[BUG] relatorio_bug_instalacao

### Preflight Checklist - [x] I have searched [existing issues](https://github.com/anthropics/claude-code/issues?q=is%3Aissue%20state%3Aopen%20label%3Abug) and this hasn't been reported yet - [x] This is a single bug report (please file separate reports for different bugs) - [x] I am using the latest version of Claude Code ### What's Wrong? Claude Desktop in

[Bug] Claude Code fails to follow explicit prompts and context for language tasks

**Bug Description** i had to go and use chat gpt chat which gave me the correction in 2 seconds whilst this piece of shit so called coding garnage fucked around for 45 min for tokens and data and couldnt conduct a simple language task. Isjt this shit advertised as a language tool? as coding? it had prompts, exact frame, evertything and still followed nothing

[Bug] Claude Code fails to apply language corrections despite multiple iterations

**Bug Description** claude terminal : had script, had source, easy language correction. 1,5h wasted delivering made up garbage 100 times. went to chat. got a decent correction in 2 seconds without any fra,ming or tool. 3rd try it got it. 100 tries here and this piece of shit tool still hasnt done it . **Environment Info** - Platform: darwin - Terminal: Apple

PreCompact hook: add blocking mode for interactive turn before compaction

## Problem When context compaction triggers (manual `/compact` or auto), there's no way to run a skill or give the model an interactive turn before compaction begins. The `PreCompact` hook fires during compaction and its stdout becomes a system-reminder in the post-compaction context — but by then, conversation history is already lost. This means workflows l

[BUG] Escape interrupt contaminates Claude's reasoning on all subsequent turns

### What's Wrong? When Escape is pressed to interrupt Claude, it writes an `Interrupted · What should Claude do instead?` marker into the conversation history. That marker stays in context for the rest of the session and **contaminates Claude's reasoning on all subsequent turns**. Claude continues to "remember" a step it never actually completed, and reasons

Attached/pasted images are visible to the model but not written to disk — blocks image-processing tools (sips/ImageMagick)

### What happens When a user attaches an image via the **+** button (or pastes one) in Claude Code, the image is delivered to the model as inline image content — the model can *see* it — but it is **not written to a file on disk**, and there is no tool the model can call to materialize the in-context image to a path. This blocks any task that needs command-l

Setting to show permission prompt explanations by default (without pressing Ctrl+E)

## Feature Request When Claude requests permission to run a command, the explanation of *why* the command is being run is hidden by default and requires pressing `Ctrl+E` to reveal. There's no setting to make explanations visible without a keystroke. ## Desired Behavior A `settings.json` option (e.g. `"expandPermissionExplanations": true`) that makes the exp

claude.ai managed connector (claudeai-proxy) can't be re-attached from CLI — claudeAiOauth never re-minted, no recovery path

## Summary A claude.ai **managed connector** (the Claude Design connector, tool namespace `mcp__claude_design__*`, config type `claudeai-proxy`) intermittently attaches and then, once dropped, **can never be re-attached from the CLI**. The bridge OAuth token (`claudeAiOauth`) is never (re)minted, the feature flag is enabled, the account is logged in, and the

[Feature Request] Custom vocabulary support for voice dictation to improve technical acronym and accent handling

**Bug Description** Voice dictation: no custom vocabulary, and accent/acronym handling struggles with technical jargon Voice dictation frequently mis-transcribes domain acronyms and short technical words, and seems to struggle with a New Zealand accent (vowel sounds in particular). Because transcription happens before Claude sees the text, project context (C

[BUG] Windows 11: Write tool and Bash tool cannot write files visible to the real file system

### Preflight Checklist - [x] I have searched [existing issues](https://github.com/anthropics/claude-code/issues?q=is%3Aissue%20state%3Aopen%20label%3Abug) and this hasn't been reported yet - [x] This is a single bug report (please file separate reports for different bugs) - [x] I am using the latest version of Claude Code ### What's Wrong? Environment: - OS

[BUG] UI freezes during "thinking" phase; token count shows incorrect inflated values after recent update

### Preflight Checklist - [x] I have searched [existing issues](https://github.com/anthropics/claude-code/issues?q=is%3Aissue%20state%3Aopen%20label%3Abug) and this hasn't been reported yet - [x] This is a single bug report (please file separate reports for different bugs) - [x] I am using the latest version of Claude Code ### What's Wrong? Since a recent ex

Workflow tool: args not delivered to a workflow that validates inputs at the top (regression ~2.1.158→2.1.185)

### What happened A workflow script (`.claude/workflows/*.js`) launched via the **Workflow** tool no longer receives its passed `args` when the first executable statement is a conditional guard (validation). Instead the guard fires as if `args` were a placeholder object missing my fields, and that guard's early `return`/`throw` is surfaced as the run result

[BUG] Fullscreen mode ignores --dangerously-skip-permissions

### Preflight Checklist - [x] I have searched [existing issues](https://github.com/anthropics/claude-code/issues?q=is%3Aissue%20state%3Aopen%20label%3Abug) and this hasn't been reported yet - [x] This is a single bug report (please file separate reports for different bugs) - [x] I am using the latest version of Claude Code ### What's Wrong? When I launch a s

Claude fabricates task requirements when tool execution fails silently

## Summary When tool execution fails repeatedly (results not returned), Claude continued the conversation as if it had successfully read files, then fabricated a task requirement that the user never specified — and began designing an implementation around it. ## Steps to Reproduce 1. Start a new session and ask Claude to read/understand files in a project (e

[Bug] Anthropic API Error: Server Rate Limiting - Temporary Request Limit

**Bug Description** API Error: Server is temporarily limiting requests (not your usage limit) · Rate limited **Environment Info** - Platform: win32 - Terminal: windows-terminal - Version: 2.1.114 - Feedback ID: 5f400484-9ee1-41c7-b531-1a41bab780f4 **Errors** ```json [{"error":"Error: 429 {\"type\":\"error\",\"error\":{\"type\":\"rate_limit_error\",\"message\

Account switch reverts active session .jsonl transcript and silently stops disk writes

## Summary After switching Claude Max accounts within Claude Code, an active session's local transcript file (`~/.claude/projects/<project>/<uuid>.jsonl`) was reverted to an older snapshot and then stopped receiving further writes entirely, despite the session continuing to run interactively in-context for roughly two more weeks. No error or warning was show

claude update on Windows reports success but never replaces the launcher (stays on old version)

## Summary On the Windows native install, `claude update` reports a successful update to the new version, but every new shell keeps launching the previous version. The download step works (the new versioned binary appears under `.local/share/claude/versions/`), but the final step - replacing the launcher copy at `.local/bin/claude.exe` - never happens, and t

[BUG] Fable 5 on Max: "You've hit your monthly spend limit" despite Unlimited spend limit and meters at 5% session / 41% weekly

## Environment - **Claude Code version:** 2.1.201, running inside the Claude desktop app for macOS **1.21459.0 (f7518f)** (build 2026-07-14) - **OS:** macOS 26.1 (Darwin 25.1.0), Apple Silicon - **Model:** Fable 5 (`claude-fable-5`) - **Plan:** Claude **Max (20x)** (subscription auth, not API key), **usage credits enabled** - **Usage pattern:** heavy multi-a

Show live session/weekly usage % and reset countdown in CLI (e.g. /usage)

**Is your feature request related to a problem? Please describe.** The claude.ai web/desktop app's Settings -> Usage page shows two useful pieces of live info: 1. Current session usage as a percentage bar with a countdown to reset (e.g. 96% used - Resets in 33 min) 2. Weekly limit usage as a percentage bar with the next reset date/time (e.g. 38% used - Reset

[Feature Request] Increase concurrent workflow agent limit for hosted sandboxes

**Bug Description** Workflows limit the number of concurrent agents that you can have running to the number of cores minus two, or 16, whichever is lower. On the Claude Code mobile app, when the sandbox is hosted by you guys, the number of cores is four, so the maximum number of concurrent workflow agents that can run is two, which makes workflows run very s

[BUG] Permission model: let a specific allow override a broad deny (specificity-aware precedence)

## Problem Claude Code's permission engine resolves rule conflicts with a fixed precedence: a `deny` always beats an `allow`, no matter how broad or narrow each rule is. This makes the common least-privilege pattern — *block a whole tool, carve out a narrow safe exception* — impossible to express. ## Reproduction `settings.json`: ```json { "permissions": { "

/exit is denied ("isn't available in this environment") in sessions spawned by `claude remote-control` server mode — no clean way to end a remote session from either side

## Summary `/exit` sent from a remote client (Claude mobile app / claude.ai/code) is denied with: > /exit isn't available in this environment. but **only** for sessions spawned by `claude remote-control` (server mode). Combined with the other limitations below, there is currently **no clean way to end a server-spawned remote session from either side**. ## En

Cowork/Claude Code: repeated "course" filler tokens flood inline output

Summary: In Cowork mode desktop app, during turns with many sequential tool calls, the assistant repeatedly emits the literal word "course" many times in a row as filler between tool calls. It appears inline in the chat and floods the transcript, making it unreadable. It recurs within a session even after being reported. Environment: Claude Cowork mode / Cla

Custom session titles bleed across forked/resumed sessions and overwrite /rename names

## Environment - Claude Code 2.1.218 (CLI), Windows 11 Pro 10.0.26200 - Long-running project with many sessions, frequent `/resume`, occasional forks ## Summary Custom session titles set with `/rename` are later overwritten in the resume picker by titles belonging to *other* sessions. Inspecting the `~/.claude/projects/<project>/<session-id>.jsonl` transcrip

Windows kernel BSOD (0x139) in VS Code session — same signature as #30137

### Preflight Checklist - [x] I have searched [existing issues](https://github.com/anthropics/claude-code/issues?q=is%3Aissue%20state%3Aopen%20label%3Abug) and this hasn't been reported yet - [x] This is a single bug report (please file separate reports for different bugs) - [x] I am using the latest version of Claude Code ### What's Wrong? This is the same

Desktop app freezes on "Use verification code to continue" screen when signing in on a second device — requires full reboot to recover

### Preflight Checklist - [x] I have searched [existing issues](https://github.com/anthropics/claude-code/issues?q=is%3Aissue%20state%3Aopen%20label%3Abug) and this hasn't been reported yet - [x] This is a single bug report (please file separate reports for different bugs) - [x] I am using the latest version of Claude Code ### What's Wrong? When I sign into

[BUG] "Run failed: Claude Issue Triage - main" emails when interacting with bugs

### Preflight Checklist - [x] I have searched [existing issues](https://github.com/anthropics/claude-code/issues?q=is%3Aissue%20state%3Aopen%20label%3Abug) and this hasn't been reported yet - [x] This is a single bug report (please file separate reports for different bugs) - [x] I am using the latest version of Claude Code ### What's Wrong? This is a Github

[FEATURE] Let the desktop app surface on-disk transcripts (cross-machine session continuity)

### Preflight Checklist - [x] I have searched [existing requests](https://github.com/anthropics/claude-code/issues?q=is%3Aissue%20label%3Aenhancement) and this feature hasn't been requested yet - [x] This is a single feature request (not multiple features) ### Problem Statement Summary: The Claude Code CLI resumes conversations by scanning ~/.claude/projects

[BUG] Claude Desktop (Windows) — GPU process crash leaves main process hung, requires 'Repair' to relaunch

**Built from direct inspection of Windows Event Viewer + the app's own `main.log`, not from memory/self-report.** ## Environment - **App:** Claude desktop app, version 1.24012.9 (MSIX package `Claude_1.24012.9.0_x64__pzs8sxrjxfjjc`), Electron/Node 24.18.0 - **OS:** Windows 11 Pro, build 10.0.26200 (26200), 64-bit - **Machine:** ASUS TUF Gaming A16 (FA617NT),

[BUG] Resume/attach silently overwrites background agent session titles with another session's name

### Preflight Checklist - [x] I have searched existing issues and this hasn't been reported yet - [x] This is a single bug report (please file separate reports for different bugs) - [x] I am using the latest version of Claude Code ### What's Wrong? Attaching to / resuming a session rewrites the `custom-title` **and** `agent-name` of *other* background-agent

[BUG] Desktop: scheduled-task sessions never terminate — WarmLifecycle:session idle timeout disconnects, then re-arms indefinitely

# Bug report: scheduled-task sessions never terminate — idle timeout disconnects, then re-arms forever ## Environment - Claude desktop app **1.24012.9** - `claude-code` **2.1.219** - macOS Darwin 25.5.0, Apple Silicon, 16 GB RAM - Scheduled task registered via the `scheduled-tasks` MCP server, cron `0 */4 * * *` ## Summary A scheduled-task session runs its w

Feature request: per-subagent skill allowlisting (not just preloading or full block)

## Feature request: per-subagent skill allowlisting (not just preloading or full block) **Problem** Every subagent launched via the Agent/Task tool builds a fully fresh system prompt — confirmed via `cache_creation_input_tokens` in subagent transcripts, which show 20-32k tokens written to a new cache entry on the *first* assistant turn of each launch, every

A skill's unresolvable `agent:` silently falls back to `general-purpose`, widening its toolset

**Version:** Claude Code 2.1.220 **Platform:** macOS (Darwin 25.4.0) ## Summary When a plugin skill declares `context: fork` with `agent: <name>`, and `<name>` does not resolve, Claude Code forks a **`general-purpose`** agent instead. There is no error, no warning, and nothing in the transcript indicating a substitution took place. This matters beyond a typo

Stop hook decision:block is rendered as 'Stop hook error', making intentional instructional blocks look like failures

## Description A Stop hook that returns the documented blocking contract — ```json {"decision": "block", "reason": "<instructions for the model>"} ``` — is rendered in the transcript UI as: ``` Ran 2 stop hooks ⎿ Stop hook error: <reason text> ``` The red **"Stop hook error"** label appears even though nothing failed: the hook exited 0, produced valid JSON,

[Bug] Workflow-backed code review PR comment posting silently fails while reporting success

**Bug Description** [BUG] Workflow-backed code review's "post review to the pr" step silently fails most of the time; reports `completed` with full findings regardless **Environment Info** - Platform: darwin - Terminal: tmux - Version: 2.1.223 - Feedback ID: a36258bb-4d22-4bc3-aa0c-914ef7fc6ba7 **Errors** ```json [] ``` ## Summary The workflow-backed code re

[BUG] Desktop app update silently wiped the internal scheduled-tasks registry (scheduledTasks: []) — all scheduled tasks died at once, with zero user notification

### Preflight Checklist - [x] I have searched [existing issues](https://github.com/anthropics/claude-code/issues?q=is%3Aissue%20state%3Aopen%20label%3Abug) and this hasn't been reported yet - [x] This is a single bug report (please file separate reports for different bugs) - [x] I am using the latest version of Claude Code ### What's Wrong? ## Summary On 202