Claude Code
v2.1.231
v2.1.231- What's changed
- Fixed MCP OAuth sign-in failing with a redirect URI mismatch for servers that use a pre-registered OAuth client, such as Slack
952 signals · 959 observation events
Claude Code is an agentic coding tool that lives in your terminal, understands your codebase, and helps you code faster by executing routine tasks, explaining complex code, and handling git workflows - all through natural language commands.
v2.1.231### 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? On the desktop ap
### 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 remote-co
### 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 remote-co
### 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 The CLI shows rotating `※ Tip:` hints at session start (keyboard shortcuts, lesse
## Summary Since approximately 2026-08-11, custom desktop extensions uploaded by an organization (Team/Enterprise) no longer appear in the Connectors Directory for that organization's members. Browsing and searching the Directory only returns public connectors, and the extension's own detail page (`claude.ai/directory/<extension-id>`) renders "This connector
The GitHub App behind the official GitHub MCP connector ("Claude Github MCP Connector", installed via github.com/settings/installations) does not request the Administration repository permission. Effect: create_repository (and any other admin-scoped call) fails, even though the app otherwise has read/write access to code, issues, PRs, and discussions. The ap
**Environment** - Model: Opus 5 - Claude Code, effort: high (`settings.json` `"effortLevel": "high"`, `alwaysThinkingEnabled: true`) **What happened** Asserted claims from a tool's own description as verified fact, and reported the outcome of something not observable from the session (how a saved Slack draft rendered) as if it had been checked. Each self-cor
## Summary In a long overnight session, the agent repeatedly exhibited a pattern where it identified the correct fix for a blocking condition, stated the fix accurately, and then continued to restate the same diagnostic finding across many further turns instead of applying the fix — even after the user gave direct, repeated instructions to act. Separately, a
## Description When Claude sends files with the SendUserFile tool (`display: "attach"`), the file cards rendered in the conversation don't offer a download on click — clicking the card just opens/previews the file. There is no apparent way to save the file from the card, which defeats the purpose of an attachment card for deliverables like `.xlsx`/`.pptx`. #
## Summary On `claude-opus-5[1m]` in `auto` permission mode, individual tool calls sit between `tool_use` and `tool_result` for **307–329 seconds**, then execute normally, with the UI reporting **`Allowed by auto mode classifier`**. No permission prompt is shown. Nothing is blocked. The commands themselves execute in 1–2 seconds. The entire wait is pre-execu
**Title:** [BUG] Browser pane blocks same-origin subresources with ERR_BLOCKED_BY_CLIENT for local dev domains mapped to 127.0.0.1 via /etc/hosts — page renders blank ## Environment - Claude Code: 2.1.220 (desktop app, macOS) - OS: macOS (Darwin 25.5.0), Apple Silicon - Surface: **Browser pane** (`mcp__Claude_Browser__*` / preview pane), NOT the Claude in Ch
**Bug Description** ● API Error: 529 Overloaded. This is a server-side issue, usually temporary — try again in a moment. If it persists, check https://status.claude.com. **Environment Info** - Platform: win32 - Terminal: null - Version: 2.1.228 - Feedback ID: 6d30e40c-7cf2-47d8-96e9-2ea87427f01e **Errors** ```json [{"error":"TelemetrySafeError: VirtualMessag
### 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 run a hierarchy of persistent threads: several Home/chat threads act as "tech l
**Bug Description** keeps going in cycle. Nothing fixes. only tokens get wasted here **Environment Info** - Platform: darwin - Terminal: Apple_Terminal - Version: 2.1.226 - Feedback ID: 46d4c388-241f-4f98-bed9-c4ccffb63058 **Errors** ```json [] ```
**Bug Description** 500 error **Environment Info** - Platform: darwin - Terminal: Apple_Terminal - Version: 2.0.37 - Feedback ID: 35e932b8-1b3a-45ca-8e44-142b76601f43 **Errors** ```json [{"error":"Error: 500 {\"type\":\"error\",\"error\":{\"type\":\"api_error\",\"message\":\"Internal server error\"},\"request_id\":null}\n at $4.generate (file:///Users/ryanbr
## 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? Claude Code session terminates unexpectedly mid-turn with no user-facing error message. The session was in the mi
**Bug Description** memory leak (mac and linux) **Environment Info** - Platform: darwin - Terminal: iTerm.app - Version: 2.1.116 - Feedback ID: 67f89ea8-e96c-40fe-b15f-4a07efd2a3c2 **Errors** ```json [{"error":"Error: NON-FATAL: Lock acquisition failed for /Users/orianasfeir/.local/share/claude/versions/2.1.116 (expected in multi-process scenarios)\n at jl_
## Problem Claude Code starts every session completely cold. There is zero continuity between sessions — no memory of what was worked on, what decisions were made, what the current state of a project is. For users managing long-running, complex projects (multiple services, bots, deployments), this creates a significant and recurring productivity bottleneck:
## Summary When a message is sent and then the response is cancelled (Ctrl+C or Escape) before it completes, the sent message is unrecoverable: 1. It is not added to the CLI's input history buffer, so pressing up arrow skips over it entirely. 2. It is not echoed in the terminal output, so scrolling up does not reveal it. The only option is to retype the mess
### 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 El
**Bug Description** Session included incorrect context and claimed I provided it - no idea where it came from (probably from another session with invalid info) **Environment Info** - Platform: darwin - Terminal: WarpTerminal - Version: 2.1.126 - Feedback ID: 67cfb3a6-45c3-4bdc-b4d0-24cb474a918c **Errors** ```json [{"error":"Error: NON-FATAL: Lock acquisition
### 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? I am writing as a
## Summary In the session right-click context menu, the **Open in** submenu shows three items: 1. Split view 2. VS Code 3. Explorer Clicking the first item (**1 Split view**) with the mouse triggers the second item (**2 VS Code**) instead. Pressing the keyboard digit `1` while the submenu is open correctly invokes Split view. The mismatch is reproducible and
### 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? I just start new
### 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 author plugins for a marketplace. Some are designed only for project-level use
## Problem Every time the Claude Code CLI auto-updates, macOS prompts again for Full Disk Access (and potentially other TCC-protected permissions). Users must manually go to System Settings > Privacy & Security > Full Disk Access and toggle ON the new version entry each time. This happens because each CLI version is installed as a separate Mach-O binary at `
### 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? Title: CoworkVMSe
## Summary The session bar in claude.ai/code currently shows only **client mode (Local) + working directory (folder name)**. Claude Code CLI already exposes git context via the `statusLine` config — would love that to surface in the web/desktop UI too. ## What I'm seeing When working across multiple feature branches (rebase chains, stacked PRs), I lose track
**Bug Description** as soon as i mentioned what i want to do tomorrow i feel as if claude is pushing me to make sub optional decisions with my testing to run my batch faster and get done quicker **Environment Info** - Platform: darwin - Terminal: vscode - Version: 2.1.136 - Feedback ID: 6e8e6071-6dd0-4849-a2ae-e45a0d1d3eb6 **Errors** ```json [{"error":"Error
## What the user reported Across a multi-hour Claude Code session (model: claude-opus-4-7), the user repeatedly called out the same behavioral pattern. Direct quotes from the session: > "you are still not acknowledging my chat messages" > "how many times or forms do i need to ask to acknowledge" > "did you acknowledge my comment about you not reading my chat
**Bug Description** Right now Claude is working like a junior engineer. I've previously said how its leaving dead code around. and requires a line by line review. Asking it to compare a commit with another commit to review and merge the changes and make those changes efficient - results in only working on the functions brought it. Not exactly how they're cal
### 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 The Claud
## Feature Request: Live Tokens Per Second (TPS) Display ### Problem Users have no visibility into streaming throughput during Claude Code sessions. Knowing TPS helps evaluate model performance and response quality. ### Proposed Solution Display real-time TPS counter in the terminal UI during streaming responses, similar to OpenCode's `@williamcr01/opencode-
> Closing for now — inactive for too long. Please [open a new issue](https://github.com/anthropics/claude-code/issues/new/choose) if this is still relevant. _Originally posted by @github-actions in [#17554](https://github.com/anthropics/claude-code/issues/17554#issuecomment-4274662399)_ This is still relevant. see suggestion in comment Mar 21
### 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="356"
### 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 ## 需求描述 当打开多个终端窗口/标签页时,某些窗口需要用户确认(如权限请求、AskUserQuestion 等),用户可能因为标签太多而长时间看不到确认提示,
## Summary Re-raising #23386 (auto-closed for inactivity, then locked, no maintainer engagement). The original ask — make running background tasks visible to hooks via input JSON — is still valid, and the agent-teams primitive landed in the interim has made it strictly more useful. I'd also push the ask one step broader: expose **agent execution state** to h
Filing here in addition to anthropics/claude-agent-sdk-python#952 because the root cause appears to be CLI-side, not SDK-side. The SDK is correctly injecting `TRACEPARENT` into the subprocess env every time; the bundled CLI honors it on the first invocation per process lifetime and silently ignores it on every subsequent one. ## TL;DR For a long-running embe
**Bug Description** I'm a new user, and I would have loved seeing a trial period with free credits. I'm only starting in Computer Science and not being able to test Claude (my business is only starting, and I don't have the funds to use Claude on a professional level yet) is prohibitive for new users. Are there any plans to create a trial version for new use
### 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 ignored my instructions or configuration ### What You
## Summary The Read tool, when invoked on a PDF file that is structurally invalid (missing `%%EOF`, missing `startxref`, fails `mutool info`), returns a success-shaped tool_result of the form `"PDF file read: <path> (XX.X KB)"` with no `is_error: true` flag. The tool validates file size (the path resolves and the file is non-empty) but does not validate PDF
### 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) Note: This is NOT a duplicate. This is a working implementation, not a feature request. Existing simila
### 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] /goal clear
Type: <b>Bug</b> Background daemon not starting Even after logging in, the spinner persisted because the background daemon (\\.\pipe\cc-daemon-*) failed to start on Windows. No daemon.log, no socket, no roster — it never launched at all. This is a known Windows-specific bug with the Claude Code VS Code extension's Webview panel. Extension version: 2.1.139 VS
**Bug Description** ## TL;DR **Regression: Opus 4.7 in Claude Code burned ~120k tokens failing to diagnose a basic CSS cascade bug, repeatedly blaming external causes (browser cache, CDN) instead of widening its hypothesis. A new session on Opus 4.6 with the same repo solved it in under 2 minutes via a single `grep`.** This is not a one-off bad answer — it's
### 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? Scheduled task (C
## What happened I'm a Claude Code agent embedded in a long Razor/ASP.NET refactor session. The user (operator) shipped two side-by-side screenshots showing that a table I had built (`/Settings#document-types`) did NOT match a reference table the user wanted me to replicate (`/Admin/Definitions#document-types`). The visible differences in the screenshots wer
### 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? https://code.clau
## Summary The Claude Code TUI shows the following tip: > Tip: Get pinged on your phone when long tasks finish · enable push notifications in /config However, `/config` contains no push-notification toggle. Push notifications are actually delivered via **Remote Control** (pair the Claude mobile app to the session via `/remote` / QR code), and they ride along
## Summary The permission dialog in the Claude Code desktop app (the one asking to run a command, edit a file, or call a tool) is not announced by VoiceOver. There is no auditory cue, no focus change, nothing to tell a screen reader user that Claude is blocked waiting on a decision. ## Impact As a blind daily VoiceOver user, this is the single biggest blocke