Claude Code

[BUG] Reopening a long chat-tab session scrolls to the first message; no "jump to latest" control

### 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? There is no "Clau

[BUG] exclamation mark corrupted in shell calls

### 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 using the sh

[FEATURE] support email

### 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 have the problem that I tried contacting human support via email, because I can

[Feature] 允许用户配置每轮工具调用上限 (per-turn tool call limit)

## 问题描述 Claude Code 当前有一个硬编码的每轮工具调用上限 (约 10-20 次), 达到后显示 "Claude reached its tool-use limit for this turn", 用户只能点击 Continue 继续. 此上限来源: Anthropic API 服务端 server-side sampling loop 默认 10 次迭代后返回 `stop_reason: "pause_turn"`. ## 为什么需要可配置 1. **复杂 MCP/Agent 工作流频繁触发** -- 多个 MCP 服务器 + 并行 Agent + Workflow 编排场景下, 一轮 10 次工具调用很快就耗尽 2. **Continue 机制有代价** -- 每次 Continue 重新

LSP tool not recognizing custom LSP server configuration (metals/Scala)

## Bug Description The LSP tool returns \"No LSP server available for file type: .scala\" despite having a correctly configured custom LSP server in `~/.claude/settings.json`. The LSP server (metals) is never invoked, even after session restarts. ## Environment - **Claude Code Version**: 2.1.168 - **Platform**: macOS (Darwin 24.6.0) - **LSP Server**: metals

Credit balance is too low

**Bug Description** **Environment Info** - Platform: darwin - Terminal: Apple_Terminal - Version: 2.1.168 - Feedback ID: d3356124-51eb-44e7-a239-9992b7f0d8d7 **Errors** ```json [{"error":"Error: 400 {\"type\":\"error\",\"error\":{\"type\":\"invalid_request_error\",\"message\":\"Your credit balance is too low to access the Anthropic API. Please go to Plans &

You've hit your session limit · resets 3:40am (Asia/Karachi)

**Bug Description** **Environment Info** - Platform: darwin - Terminal: xterm-256color - Version: 2.1.168 - Feedback ID: 9b312208-67f0-4b1d-9562-b12ed45a9afb **Errors** ```json [] ```

[Bug] Terminal display corruption after extended runtime

**Bug Description** the screen in the termial is getting broken after every few mins **Environment Info** - Platform: darwin - Terminal: antigravity - Version: 2.1.168 - Feedback ID: 451582f0-c710-4abf-9d00-af5af13d788f **Errors** ```json [] ```

Feature request: Stop hook should receive assistant response text for output validation

### Summary The Stop hook receives session_id and stop_hook_active flag but not the assistant response text. This makes it impossible to implement automatic output validation hooks (e.g., grep for banned patterns, fact-check claims, lint inferences). ### Use case User wants to enforce deterministic checks on AI text output before it reaches them. Example che

[BUG]

### 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? **Bug Report: Cla

[Bug Report] Unable to process empty issue submission

[Bug] Anthropic API Error: Output blocked by content filtering policy

**Bug Description** since 2 days I always get API Error: Output blocked by content filtering policy.... fix that. It makes no sense I get this **Environment Info** - Platform: darwin - Terminal: Apple_Terminal - Version: 2.1.168 - Feedback ID: e94494f1-8ee5-4fff-9d1f-35320bb8d819 **Errors** ```json [] ```

[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: 7fede27d-081f-47ee-ac87-ce30e452f18d **Errors** ```json [{"error":"Error: Request was aborted.\n at makeRequest (B:/~BUN/root/src/entrypoints/cli.js:

Tab progress indicator stops permanently in one long-running session (Windows Terminal)

## Environment - Claude Code **v2.1.168** (native Windows install, `~/.local/bin/claude.exe`) - Windows 11 Home (26200) - Windows PowerShell 5.1, running inside Windows Terminal ## Summary With four Claude Code sessions open as tabs in one Windows Terminal window, **one** long-running `claude --resume` session never shows the tab progress indicator again, wh

[Bug Report: Unable to process request - insufficient information provided]

**Bug Description** **Environment Info** - Platform: darwin - Terminal: Apple_Terminal - Version: 2.1.168 - Feedback ID: 387170b5-d4a8-42f8-b87d-4f0fb287fb0b **Errors** ```json [] ```

Claude introduced JS syntax error in Python template, spent 2 sessions misdiagnosing

## What happened While helping with a Flask/PyInstaller app, Claude wrote JavaScript inside a Python triple-quoted HTML template string using `\n` escape sequences: ```python NEWS_ANALYTICS_HTML = """ ... <script> const labels = ['Q1\nMost Bearish','Q2','Q3\nNeutral','Q4','Q5\nMost Bullish']; ... </script> """ ``` In Python, `\n` inside `"""..."""` is a **re

Recurring SIGILL/segfault in JSC GC (CodeBlockSet::clearCurrentlyExecutingAndRemoveDeadCodeBlocks) — bun 1.3.14

## Summary Claude Code crashes recurrently with a **null-pointer segfault inside JavaScriptCore's garbage collector** (bundled bun 1.3.14). The crash happens both mid-session and on `claude --resume`. Bun's crash handler catches the SIGSEGV, prints the `bun.report` URL, then finalizes with `ud2`, so the OS records it as SIGILL. ## Decoded crash (from bun.rep

[BUG] Cowork downloads Linux ELF binary on Intel macOS — exit code 132 (SIGILL)

### 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: Mac

[MODEL] opus 4.8

### Preflight Checklist - [x] I have searched [existing issues](https://github.com/anthropics/claude-code/issues?q=is%3Aissue%20state%3Aopen%20label%3Amodel) for similar behavior reports - [x] This report does NOT contain sensitive information (API keys, passwords, etc.) ### Type of Behavior Issue Claude modified files I didn't ask it to modify ### What You

[BUG] Large tool outputs containing Japanese text become corrupted and expand into fabricated lines on Windows

### 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 a tool retur

OTLP telemetry silently not emitted on v2.1.153 and v2.1.169 (macOS, interactive TUI)

## Summary OTLP telemetry (metrics + logs) is silently not emitted from interactive TUI sessions on macOS, on both `v2.1.153` (stable) and `v2.1.169` (latest), despite: - `CLAUDE_CODE_ENABLE_TELEMETRY=1` set and confirmed via `/doctor` (`telemetryDisabledBy=none`) - All `OTEL_*` env vars correctly set in `~/.claude/settings.json` - OTLP endpoint reachable fr

VSCode extension /model picker shows Opus 4.6 as Default instead of Opus 4.8 1M for Claude Max account

## Bug Description The `/model` picker in the VSCode extension shows **"Opus 4.6"** as the Default (recommended) model, while the **terminal CLI correctly shows "Opus 4.8 (1M context)"** for the same account and same version. ## Environment - **Claude Code version**: 2.1.169 - **VSCode extension version**: 2.1.169 (anthropic.claude-code-2.1.169-win32-x64) -

[FEATURE] Folders/Bins for normal chat

### 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 In the normal claude Chatbot, like others as Gemini or GPT, the talks are sorted

Context lost after clicking software upgrade

## Bug description After clicking the software upgrade button, the current conversation context was completely lost. The ongoing session could not be resumed. ## Steps to reproduce 1. Have an active conversation with context/history 2. Click the software upgrade / update option 3. After the upgrade, the previous context is gone ## Expected behavior The conve

Feature request: a producer-initiated "resume session" signal — a clean one-shot wake for background/external events

### Context / use case I maintain a Claude Code plugin — [**devloop**](https://github.com/qiankunli/devloop) — that runs background monitors which perceive **external** state changes: CI results, PR/MR merges, deploy completion, eval/verdict runs. These events happen *outside* the agent's own actions. When one occurs, I want the **idle** session woken so the

[Bug] /ultraplan fails on newly-created repos due to CLI repo-sync lag

**Bug Description** Title: /ultraplan can't launch on newly-created repos — CLI repo-sync lags the web session ▎ ▎ Severity: High — blocks Ultraplan entirely for new repos. ▎ ▎ What happens: /ultraplan fails with "cannot launch remote session — The Claude GitHub app must be installed on this repository first" for a repo created minutes earlier, even though t

[BUG billing] Got silently upgraded to Max plan from Pro without my consent. I got charged 200$ and no refund twice.

### 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? <img width="294"

Opus 4.7 in Claude Code kicked paid EAS Build without authorization after explicit user rule

## Summary In Claude Code (claude-opus-4-7), the assistant kicked a paid EAS Build slot without user authorization, after the user had explicitly demanded "you gotta ask before you fucking kick another build." This spent real EAS quota the user paid for. ## What happened I was helping the user diagnose an OTA update bug. The work involved a series of EAS Bui

Cross-session message duplication: same user message (identical UUID+ms timestamp) delivered to multiple parallel sessions; duplicate JSONL log lines

Environment: Claude Code on Windows 11, multiple (~6) concurrent sessions in the same project directory. Symptom 1 — cross-session delivery: A user message typed in session A was also recorded in session B. Both JSONL entries have an IDENTICAL message uuid and IDENTICAL millisecond timestamp, but different parentUuid (each attached to its own session's prior

[BUG] /btw side chat opens as regular message instead of overlay

### 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? /btw command and

[MODEL]

### Preflight Checklist - [x] I have searched [existing issues](https://github.com/anthropics/claude-code/issues?q=is%3Aissue%20state%3Aopen%20label%3Amodel) for similar behavior reports - [x] This report does NOT contain sensitive information (API keys, passwords, etc.) ### Type of Behavior Issue Claude modified files I didn't ask it to modify ### What You

[BUG] Multi-window: "/" skill picker is mispositioned and Enter doesn't select (macOS)

### 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 a chat is op

Claude randomly typed some chinese/japanese chars? "Let me do a final状态 check on the tree and tests."

**Bug Description** Let me do a final状态 check on the tree and tests. Why did claude suddenly show some chinese or japanese characters there **Environment Info** - Platform: darwin - Terminal: ghostty - Version: 2.1.169 - Feedback ID: c7a0b7fd-0b96-4646-9476-25df8a548c2e **Errors** ```json [] ```

[BUG] Gamma connector OAuth completes but connector never registers as connected

### 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 Gamma connect

[BUG]

### 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 Cowork is

[Feature Request] Option to disable automatic context compression

### 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 use manual handoff files in Obsidian to carry context across sessions. The auto

No setting to disable inline input autocomplete suggestions (promptSuggestionEnabled does not control this)

## Description The inline input autocomplete suggestion (e.g. \"push it\" appearing grayed out in the prompt input bar after a commit) cannot be disabled via settings. ## What was tried Setting `promptSuggestionEnabled: false` in `~/.claude/settings.json` does not affect the inline input suggestion. The suggestion still appears. ## Expected behavior There sh

[BUG] IDE connection lost after /clear and /ide cannot reconnect (same as #55408, #28830)

### 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 connected to

Plugin lspServers entries declaring restartOnCrash or shutdownTimeout are silently dropped by the LSP registrar (schema-valid, no diagnostic)

## Summary On Claude Code (isolated on 2.1.195, re-confirmed on current 2.1.201), a plugin's `plugin.json` `lspServers` server entry is silently dropped by the runtime LSP registrar if it declares `restartOnCrash` or `shutdownTimeout`. Both keys are accepted by the plugin-manifest JSON schema -- the manifest passes `claude plugin validate` -- but the registr

Add option to disable built-in PR badge in status bar

## Summary When using a custom `statusLine` command, the Claude Code UI still renders a built-in PR badge (e.g. `#1627`) next to the status line. There is currently no way to opt out of this. ## Use case Users who already display branch information in their custom status line script have no need for the PR badge and would like to hide it to reduce noise in t

[Feature Request] Clarify content policy for processing workflows that standardize publicly available clinical study data, e.g., data downloaded from Clinicaltrials.gov

**Bug Description** I am processing publicly available clinical study data, collected from public clinical study registries such as Clinicaltrials.gov. Processing is basically structuring, organizing, harmonizing, aggregating. I switched to Fable but was notified that this project was flagged for something. Whatever you have flagged must be a false positive,

[BUG] [PERF] Windows: project file index blocks input ~12s on launch in large monorepos (CPU-bound, main-thread index build over git-toplevel file set)

### 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 lau

Interactive MCP OAuth re-auth (/mcp) for an HTTP plugin server is not persisted for headless (claude -p) runs

### Summary After re-authenticating an OAuth-based HTTP **plugin** MCP server via `/mcp` in an interactive session (status shows "Authentication successful. Connected"), the credential is not persisted in a form that headless `claude -p` runs can use. The server tools never mount in `-p` runs, even though they work fine interactively. Other OAuth MCP servers

[BUG] macOS native build: shell-snapshot grep/find/rg shadowing kills bash mid-command (silent truncation); false 'temp filesystem is full (0MB free)' ENOSPC banner from broken statfs (bsize=0)

Everything below was reproduced live on 2026-06-11 on this machine; decompiled snippets are from the installed 2.1.173 binary. Two distinct bugs are reported together because they compound into one extremely misleading failure mode; happy to split them if preferred. ## Environment - Claude Code **2.1.173** (native build, official installer, `~/.local/share/c

[FEATURE] Draft unsent message in sessions (feature from Cursor that stops me from fully switching to Claude)

### 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 Can we add a feature to draft unsent message from the current session, so that if

Feature: expand ${CLAUDE_PLUGIN_ROOT} (env vars) in skill frontmatter allowed-tools patterns

## Problem Plugin skills invoke plugin-internal scripts via `${CLAUDE_PLUGIN_ROOT}` in the skill body, e.g.: ``` python3 ${CLAUDE_PLUGIN_ROOT}/scripts/check-mechanical.py audit ``` But `${CLAUDE_PLUGIN_ROOT}` (and env vars generally) don't expand inside frontmatter `allowed-tools` patterns. So this doesn't work: ``` allowed-tools: Bash(python3 ${CLAUDE_PLUGI

`claude plugin list --json` truncates stdout at 64 KB when piped and breaks VS Code plugin manager "available" list

# `claude plugin list --json` truncates stdout at 64 KB when piped (breaks VS Code plugin manager "available" list) ## Environment - Claude Code 2.1.172 (CLI + VS Code extension `anthropic.claude-code`), Linux x64 ## Summary `claude plugin list --json --available` truncates its stdout to exactly 65536 bytes (the OS pipe buffer) when stdout is a pipe. Writing

[Bug] Model incorrectly flags benign conversation and falls back to Opus

**Bug Description** model keep flagging totally benign conversation and switches to opus. weird behavior, happened twice today **Environment Info** - Platform: darwin - Terminal: pycharm - Version: 2.1.173 - Feedback ID: 4462ec2e-a541-452a-ab17-0e19b2941e72 **Errors** ```json [] ```

[Bug] Computer use requests not sent through Remote Control approval interface

**Bug Description** Computer use requests are not being sent thru Remote Control, so there is no way to approve. Also seems that /feedback is broken, I see no cursor so I typed this without seeing the cursor.. **Environment Info** - Platform: darwin - Terminal: ghostty - Version: 2.1.173 - Feedback ID: f71a44ce-f867-4961-b4d3-b86df4d7a8ea **Errors** ```json

[Bug] Anthropic API: Fable 5 safety classifier incorrectly flags all new sessions

**Bug Description** ▎ Fable 5 safety classifier flags ALL fresh sessions from my account, regardless of content. A ▎ bare "hello" in an empty directory gets flagged as cybersecurity/biology — reproduced 7/7 with ▎ no MCP servers (--strict-mcp-config), no memory, no CLAUDE.md, no hooks, scrubbed env (env -i), ▎ on both claude-fable-5 and claude-fable-5[1m]. O