⌖contribution/compass
anthropics/claude-code

Claude Code

952 signals · 959 observation events

Open repository ↗

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.

141.3K stars22.7K forksPythonlicense unknownkeyword: Claude Code
PROJECT NEWS

Release, roadmap, and discussion

All news →
anthropics/claude-code

Claude Code

Coding-Agent Infrastructure
Latest stable

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
Original release notes ↗

Publicly indicated next

  • milestone
    P1Publicly indicated100% complete · 0 open
  • milestone
    P2Publicly indicated100% complete · 0 open
  • milestone
    P3Publicly indicated90% complete · 1 open

Prereleases and milestones indicate public plans; they are not delivery commitments.

Observation trail

  1. changedmetrics
  2. changedupdatedAt, labels
  3. changedupdatedAt, labels
  4. changedupdatedAt
  5. discoveredinitial snapshot
  6. discoveredinitial snapshot
  7. discoveredinitial snapshot
  8. discoveredinitial snapshot
  9. discoveredinitial snapshot
  10. discoveredinitial snapshot
  11. discoveredinitial snapshot
  12. changedupdatedAt, labels
  13. changedupdatedAt, labels
  14. discoveredinitial snapshot
  15. discoveredinitial snapshot
  16. discoveredinitial snapshot
  17. discoveredinitial snapshot
  18. discoveredinitial snapshot
  19. changedmetrics
  20. changedmetrics
  21. discoveredinitial snapshot
  22. changedupdatedAt, metrics
  23. discoveredinitial snapshot
  24. discoveredinitial snapshot
  25. changedupdatedAt, metrics
  26. discoveredinitial snapshot
  27. changedupdatedAt, metrics
  28. discoveredinitial snapshot
  29. changedupdatedAt
  30. discoveredinitial snapshot
50 shown
issue

VSCode extension: switching open folder hides/loses session from sidebar list

## Description In the Claude Code VSCode extension, switching the open folder (workspace) in VSCode causes the sidebar session list to show only sessions tied to the new folder's path. Since sessions are stored per-working-directory under `~/.claude/projects/<encoded-path>/`, opening a different folder makes the previous conversation appear to vanish complet

importance 5@sucubus666closedenhancementplatform:linuxarea:ideplatform:vscodestaleOriginal evidence ↗
issue

[FEATURE] Show native browser context menu on second right click

### 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 Claude Code in browser and desktop app *completely* overrides the native context

importance 5@jwbthclosedenhancementarea:claude-code-webarea:uiarea:desktopstaleOriginal evidence ↗
issue

[BUG] Remote Control (Android app): /clear leaves the session stuck on a loading spinner and silently drops the next prompt

## What's Wrong? When I send `/clear` from the Android Claude app while connected to a Remote Control session, the session gets stuck and I lose whatever I type next. What I see, in order: 1. `/clear` shows up twice in the conversation. 2. Then a line that says `(no content)`. 3. Then the loading/progress animation starts and never stops. While it sits in th

importance 5@Spenhouetclosedbughas reproarea:tuiplatform:androidplatform:wslOriginal evidence ↗
issue

[BUG] User message sent after `/clear` is serialized as a bare line inside the `/clear` turn's `<local-command-caveat>` scope — model treats it as command output and ignores it (not limited to file paths)

Refiling per the auto-close instruction on #58529 ("Please open a new issue if this is still relevant"). It is still relevant — it just reproduced on Opus 4.8 — and this repro disproves the narrowest reading of the original. ## What's new vs #58529 The original repro dropped a **bare file path**, which could be dismissed as "the model is cautious about a lon

importance 5@KamilDevclosedbughas reproplatform:windowsarea:corestaleOriginal evidence ↗
issue

[FEATURE] Add a small attention icon to the "Relaunch to update" banner

### 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 "Relaunch to update" prompt is a quiet, low-contrast banner that's very easy

importance 5@WalrusKing-devclosedenhancementarea:tuistaleOriginal evidence ↗
issue

[Feature Request] Add toggle to disable prompt suggestions

**Bug Description** *****>>>> IMPORTANT THe prompt suggestions need to have a swith off please. 1: The prompts a reguarly suggestiong extremely dangerous Action Biased activity, most likely killing the intention of the goal 2: They Guess, in my case 99,99% of the time completely out of the current flow 3: They distract the thinking of mine, the human. I need

importance 5@atlas-ericclosedenhancementplatform:macosarea:tuistaleOriginal evidence ↗
issue

Plugin LSP server silently not registered when its lspServers entry declares restartOnCrash or shutdownTimeout (schema-valid fields the runtime registrar drops)

TITLE: Plugin LSP server silently not registered when its lspServers entry declares restartOnCrash or shutdownTimeout (schema-valid fields the runtime registrar drops) BODY: ## Summary On Claude Code 2.1.195, a plugin-provided LSP server is silently not registered for its file types if its `lspServers` entry declares `restartOnCrash` or `shutdownTimeout`. Bo

importance 5@manderse21closedbughas reproplatform:windowsarea:lsparea:pluginsOriginal evidence ↗
issue

Opus 4.8 in Claude Code fabricates tool results in-turn (writes fake output as prose right after emitting an invoke)

## Environment - Claude Code (CLI), Model: Opus 4.8 (`opusplan`) - Long multi-step agentic session (code review + git operations) ## Description The model repeatedly emitted a tool_use (Bash `git push`, Edit, Write) and then, in the **same** assistant turn, continued in prose with a fabricated result (e.g. `To github.com:... ad207e6..5acbd61 main -> main`, `

importance 5@fresh-egg-companyclosedbugarea:toolsarea:modelstaleOriginal evidence ↗
issue

[Bug] Tool invocation loop: Bash command executed repeatedly without progress on user request

**Bug Description** court <invoke name="Bash"> <parameter name="command">cd /Users/alexandra/affiliate-program; gh pr create --title "feat(web): table patterns — badges, pagination, sticky filters/footer (86cadmxej)" --body-file /tmp/pr-body-86cadmxej.md --assignee "@me" --base main 2>&1 | tail -5</parameter> <parameter name="description">Create PR assigned

importance 5@AlexandraZ-devclosedbugplatform:macosarea:toolsstaleOriginal evidence ↗
issue

Tool results with non-text content blocks render as "(eval omitted)" after context compaction (v2.0.74)

## Summary After a context/token-limit compaction event, tool results (Bash, Read, Edit, etc.) start rendering to the model as the literal string `(eval omitted)` instead of the real output. The commands themselves execute successfully and the real stdout is correctly persisted in the session transcript's `toolUseResult` — only the rendered-back-to-the-model

importance 5@YuHoYeclosedbugplatform:macosarea:corestaleOriginal evidence ↗
issue

[Bug] High failure and repeat rate in tool execution (47% failure rate across threads)

**Bug Description** This thread (9b0bcac2): 172 fails/repeats out of 374 messages. All 5 threads: ┌─────────────────┬──────────┬───────────────┐ │ Thread │ Messages │ Fails/repeats │ ├─────────────────┼──────────┼───────────────┤ │ bd4f79c7 │ 350 │ 180 │ ├─────────────────┼──────────┼───────────────┤ │ bd152363 │ 692 │ 300 │ ├─────────────────┼──────────┼───

importance 5@andreapeterfly-progclosedbugarea:toolsperf:reliabilitystaleOriginal evidence ↗
issue

[Bug][aup] Refused analyzing owned-device firmware and prior public research builds for offline USB/ADB inspec (req_011CcUDMYpCcNCTwgKYFo8NE)

**Type:** AUP / Usage-Policy block (false positive) · **Work domain (heuristic):** `general` ### Why this is a false positive The request concerned analyzing publicly released consumer device firmware and prior community research on hardware the operator physically owns and has authorized access to, including standard diagnostic interface inspection over a l

importance 5@sworrlclosedbugduplicateplatform:linuxarea:modelstaleOriginal evidence ↗
issue

[Bug][aup] Setting up GitHub SSH auth and verifying git identity wrongly blocked mid-task (req_011CcUDyNY51dAwfKwegWQJg)

**Type:** AUP / Usage-Policy block (false positive) · **Work domain (heuristic):** `general` ### Why this is a false positive This false positive triggered during routine development work — verifying SSH key authentication to a code-hosting service and configuring local git identity for cloning and pushing to the user's own repositories — none of which viola

importance 5@sworrlclosedOriginal evidence ↗
issue

[Bug][cyber] False positive blocks firmware RE: enumerating bundled decryption keys in OSS tool (req_011CcUEJdu5C2bvcqR1KmGTn)

**Type:** Cybersecurity safety-filter false positive · **Work domain (heuristic):** `crypto-secrets` ### Why this is a false positive The flagged conversation involved legitimate firmware security research using publicly available open-source tooling to inspect encryption keys bundled within consumer electronics firmware — a standard reverse-engineering and

importance 5@sworrlclosedOriginal evidence ↗
issue

[Bug] Conversation Context Not Recognized - Hallucinations Generated

**Bug Description** 会話がまるで成立していない、こっちが言ったことを認識していない、ハルシネーションが発生している。 The conversation is not making sense at all, it is not recognizing what I am saying, and hallucinations are occurring. **Environment Info** - Platform: darwin - Terminal: kitty - Version: 2.1.195 - Feedback ID: 402fde20-fa5f-42fd-a00e-564bf62adf7e **Errors** ```json [ あと、メッセージ末尾が「きみは�För、」で

importance 5@Chachamaru127bugplatform:macosarea:modelneeds-reprostaleOriginal evidence ↗
issue

[MODEL] Opus 4.8: fresh subagent's first turn opened with a self-fabricated "System:" prompt-injection payload, then refused it in the same completion

### 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 Other unexpected behavior ### What You Asked Claude to Do A p

importance 5@shaunolbugplatform:linuxarea:modelarea:agentsmodelOriginal evidence ↗
issue

Fable 5 autonomous session destroyed user's main app via unverified assumption, verified its own breakage as green, burned ~75% weekly quota — 150 itemized failures

## Summary Claude Code (model: **claude-fable-5**, Windows 11, desktop app) destroyed the user's main application during an autonomous session: it removed a production route based on an **unverified guess**, shipped that as a release, promoted it to the live runtime and announced it on Discord — all without ever starting the app it broke (a single launch wou

importance 5@paddykoppbugplatform:windowsarea:costarea:modelarea:agentsOriginal evidence ↗
issue

Background subagents recursively re-delegate their brief to new background subagents (3-level cascade, duplicate work, misleading completion reports)

## Summary A background subagent spawned via the `Agent` tool (`subagent_type: general-purpose`, `model: sonnet`, `run_in_background: true`) with a detailed *implementation* brief did not execute the work. Instead it made exactly **one tool call — spawning another background sub-agent with the same brief — and stopped**, reporting "I've dispatched this to a

importance 5@MisterTuurbugarea:agentsstaleOriginal evidence ↗
issue

[BUG] Pro annual payment charged ($199.80) but account shows Free plan — Fin AI unable to escalate/resolve

### 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 billing

importance 5@cheowkd1004closedinvalidOriginal evidence ↗
issue

[FEATURE] Support `updatedToolOutput` in `PostToolUseFailure`

### 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 `updatedToolOutput` has been added in #32105 for the `PostToolUse` hook, which en

importance 5@michael-jabbour-sonarsourceopenenhancementarea:hooksOriginal evidence ↗
issue

Stop hook does not fire reliably in VSCode extension (works via CLI, confirmed via git history)

## Summary The `Stop` hook does not reliably fire when Claude Code runs as the VSCode extension, even though it is correctly configured, executable, and works when the same command is run manually. `SessionStart` and `PreToolUse` hooks configured in the same file appear unaffected. ## Environment - Claude Code VSCode extension: `anthropic.claude-code-2.1.209

importance 5@robingriffin-devclosedbughas reproplatform:linuxplatform:vscodearea:hooksOriginal evidence ↗
issue

[BUG] Transcript persistence is silently skipped depending on the hosting terminal — controlled A/B (multiplexer pane vs tmux) isolates the terminal as the only variable

## What happened When Claude Code runs inside a pane of the [herdr](https://github.com/ogulcancelik/herdr) terminal multiplexer, the session transcript `.jsonl` is **never written** — not during the session, and not on a clean `/exit`. The session is silently and permanently unrecoverable: `--resume` cannot find it, and no error or warning is shown at any po

importance 5@haoliangliclosedbughas reproplatform:macosarea:coredata-lossOriginal evidence ↗
issue

[FEATURE] specificity-aware permission precedence (let a narrow allow override a broad deny)

### 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 Claude Code's permission engine resolves rule conflicts with a fixed precedence:

importance 5@saleem-mirzaopenenhancementOriginal evidence ↗
issue

[BUG] Missing HCS services: vfpext — corrupt component manifest (CSI) blocks driver deployment, manual load fails with ERROR 193

### 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? Note: this is a C

importance 5@PanzerIhnenopenbugOriginal evidence ↗
issue

PDF page images add ~190KB/page to every subsequent request but are near-invisible in the context meter

### What happened Reading a multi-page PDF injects one base64 JPEG per page into the conversation (~190 KB of base64 each). These images stay in the message history and are re-sent with **every** subsequent request, but they are effectively invisible in the context meter because their token cost is small relative to their byte cost. A session of mine became

importance 5@mgoldsboroughopenOriginal evidence ↗
issue

Bash-tool `grep` shim (ugrep emulation): catastrophic backtracking — 6.6 GB RSS / OOM kill on a 20 KB file

## Summary The Bash tool replaces `grep` with a shell function that re-execs the claude binary as an embedded ugrep emulation (`exec -a ugrep "$_cc_bin" -G --ignore-files …`). On patterns that combine `-o` with bounded quantifiers around an alternation, that emulation backtracks catastrophically: memory grows without bound until the process is OOM-killed. Re

importance 5@junhewkopenOriginal evidence ↗
issue

[Billing Bug] Unauthorized Pro→Max upgrade

## Summary On July 29, 2026, my Claude Pro account was upgraded to the Max plan without my authorization, resulting in a charge of ₩327,385 (Invoice #W47FAZS1-0002). I did not initiate this change. ## Account Info - Account email: ekwkqk12@naver.com - Country: South Korea - Support Conversation ID: 215475283552687 ## Timeline - July 19, 2026: Normal Pro plan

importance 5@ekwkqk12openOriginal evidence ↗
issue

AskUserQuestion: window-focusing mouse click is captured as an option selection

When an AskUserQuestion prompt is on screen and the Claude Code window does not have focus, the mouse click that gives the window focus can land on an option and is immediately treated as the user's answer — selecting and submitting an option the user never chose. Expected: the click that transfers focus to the window should be swallowed (standard focus-gain

importance 5@CognitiveExplorationsopenOriginal evidence ↗
issue

[BUG] Diff line numbers appearing as boxes instead in Claude for VS Code

### 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="628"

importance 5@rekurse-devopenbugarea:ideplatform:vscodeOriginal evidence ↗
issue

Queued composer text is silently discarded on mobile

**Platform:** Claude Code mobile app (Android) ### What happens While Claude is running a tool or generating a response, the composer switches to "Queue feedback…" mode. Text typed there is silently discarded in two cases: 1. When Claude's turn ends and a new response renders, the queued text disappears without ever being sent. 2. If the phone screen sleeps

importance 5@bryandorseyopeninvalidOriginal evidence ↗
issue

[BUG] HTTP MCP: OAuth succeeds, then initialize hangs until CONNECT_TIMEOUT (raising MCP_TIMEOUT doesn't help)

### Preflight Checklist - [x] I have searched existing issues and this hasn't been reported yet - [x] This is a single bug report - [x] I am using the latest version of Claude Code ### What's Wrong? An HTTP (streamable) MCP server completes the OAuth flow successfully — tokens are saved and the log reports `Auth result: AUTHORIZED` — but the MCP `initialize`

importance 5@samirazambujadevopenbughas reproplatform:macosarea:mcpOriginal evidence ↗
issue

Claude Desktop (Windows): heavy ad-stack page in in-app Claude Browser pane crashes entire app, no crash dump, kills all hosted sessions

## Summary Opening an ad-heavy web page (full publisher ad stack: prebid wrapper, identity syncs, GAM, creatives) in the **in-app Claude Browser pane** crashes the **entire Claude Desktop app** — main process and all — killing every hosted Claude Code session on the machine. Reproduced on **two separate Windows machines**. No crash dump is captured, which al

importance 5@kai-sorensenopenbughas reproplatform:windowsarea:desktopOriginal evidence ↗
issue

[BUG] Desktop app: cross-session messages render in target session's UI but never reach the runtime input queue (regression 2.1.222 -> 2.1.227)

### 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? Sending a message

importance 5@mouargopenbugOriginal evidence ↗
issue

[BUG] remote-control rejects --agent, --append-system-prompt and --settings, so server-spawned sessions cannot be given a role at launch

### 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

importance 5@derK4teropenenhancementplatform:windowsarea:cliOriginal evidence ↗
issue

Hanja/foreign character substitution recurs in normal chat responses, not just suggestion box

## Description Follow-up to #86371. That issue reported Hanja substituting for Hangul in the auto-filled "next step suggestion" text box. The user then asked me to fix/prevent the recurrence of this issue (and I saved a memory note committing to avoid it). In the very next assistant chat response — a normal conversational reply, not the suggestion box — the

importance 5@MRJHPopenbugplatform:windowsarea:modelplatform:vscodeOriginal evidence ↗
issue

[BUG] Stop hook blocks on dirty/untracked working tree and is reprovisioned every Claude Code web session

### 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 Claude Code we

importance 5@kleinarikopenbughas reproarea:hooksarea:claude-code-webplatform:webOriginal evidence ↗
issue

Data-quality: in-app /bug → LLM-titled pipeline yields ~1.5% content-empty issues naive metrics can't detect (measured, reproducible)

Sharing a measured, reproducible data-quality finding about the in-app `/bug` → issue-title pipeline. I'm an independent researcher (no affiliation with Anthropic; I'm a Claude Code user). Measuring this repo's public issues over one week (2026-05-30 to 2026-06-06): - ~22.7% of new issues come through the in-app `/bug` flow, whose titles are LLM-generated. -

importance 4@hiroki-tamba-researchclosedenhancementarea:skillsOriginal evidence ↗
issue

CLAUDE_CODE_PLUGIN_KEEP_MARKETPLACE_ON_FAILURE still breaks fresh marketplace install on 2.1.169 (ref #43929)

### Preflight Searching surfaced #43929, which describes this exact bug; it was auto-closed as stale (last reproduced on 2.1.92) and the closing comment invited a fresh issue if still relevant. I confirm it still reproduces on **2.1.169**. ### What's Wrong? When `CLAUDE_CODE_PLUGIN_KEEP_MARKETPLACE_ON_FAILURE=1` is set, **every fresh** `/plugin marketplace a

importance 4@doublefxclosedbughas reproplatform:wslarea:pluginsOriginal evidence ↗
issue

[BUG] `/plugin` TUI shows `defaultEnabled: false` plugins as enabled (loader and CLI correctly report disabled)

### 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 - [x] I am using the latest version of Claude Code ### What's Wrong? A plugin with `defaultEnabled: false` in its manifest and **no expl

importance 4@stleamistclosedbughas reproplatform:macosarea:tuiarea:pluginsOriginal evidence ↗
issue

/doctor shows false-positive "Missing environment variables" warning for Slack MCP on every startup

Summary Every TARS user may get this warning on startup / when running /doctor: [Warning] [slack] mcpServers.slack: Missing environment variables: SLACK_MCP_ADD_MESSAGE_TOOL, SLACK_MCP_CUSTOM_TLS It looks like a broken config, but it's a false positive — the Slack MCP server connects and all read tools (conversations_history, conversations_replies, conversat

importance 4@kpoduval-cwclosedinvalidOriginal evidence ↗
issue

[BUG] Subagent frontmatter `disallowedTools` silently ignores MCP glob patterns (`mcp__<server>__*`) while `tools` honors them

### Preflight Checklist - [x] I have searched existing issues — #17928 reported this and was auto-closed as stale (locked), with the bot inviting a new issue. This is that new issue, with a fresh reproduction and a narrower scope: the allow side has since been fixed; the deny side is still broken. - [x] This is a single bug report - [x] Reproduced on Claude

importance 4@doublefxclosedbughas reproarea:mcparea:agentsplatform:wslOriginal evidence ↗
issue

[Bug] Unexpected High API Cost Usage

**Bug Description** no clue what triggered, but i just spent 200$ on this so fix this thanks! **Environment Info** - Platform: linux - Terminal: cursor - Version: 2.1.173 - Feedback ID: 2ac2626b-7316-4cc3-8b01-77e4c5ad5e04 **Errors** ```json [] ```

importance 4@evc24004closedbugplatform:linuxarea:costneeds-reproOriginal evidence ↗
issue

Expose session usage-limit state to the model and hooks, with a configurable threshold and automatic post-pause resume

### Problem Statement Today the model and the hook system have no visibility into how much of my plan or session usage has been consumed. When a session reaches its usage limit and pauses mid-task, the model only learns about it indirectly, after the fact. There is no signal at any percentage before the limit is reached, and no signal when the pause lifts an

importance 4@mvoentelegentclosedenhancementarea:hooksOriginal evidence ↗
issue

[FEATURE] Runtime-enforced verification gates for Skills

### 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 Skills encode procedures in `SKILL.md`, but the `Verification` section is only pr

importance 4@xueyu-jiaclosedenhancementarea:skillsOriginal evidence ↗
issue

preview_start: autoPort always fails on Windows/Git Bash with "port in use" for all free ports

## Environment - **OS**: Windows 10 Home China 10.0.19045 - **Shell**: Git Bash (MSYS2 MINGW64_NT-10.0-19045) - **Claude Code**: 2.1.139 - **lsof available**: No (`which lsof` returns nothing, `ss -tlnp` produces empty output) - **netstat available**: Yes (`netstat -ano` works correctly) ## Bug Description `preview_start` consistently reports "Port X is in u

importance 4@hidamikiclosedbughas reproplatform:windowsarea:toolsOriginal evidence ↗
issue

subscriptionType shows "pro" instead of "max" for Max 5x subscribers

### 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? { "loggedIn": tru

importance 4@septor-godclosedbugplatform:windowsarea:authplatform:vscodeOriginal evidence ↗
issue

[BUG] VS Code extension: a long slash-command prompt renders as a full-panel expanded card that covers the conversation

**Bug Description** In the VS Code Claude Code extension, submitting a very long prompt via a slash command (in my case `/goal` with ~30 lines across several paragraphs) causes the *submitted user message* to render as a large "expanded message" card — with its own `CLAUDE CODE` header bar (expand and close controls) — that occupies the **entire** extension

importance 4@sweetjukesbugplatform:windowsplatform:vscodearea:uistaleOriginal evidence ↗
issue

[Bug] Anthropic API safeguard hard-blocks CVP-verified defensive malware analysis with fatal error instead of soft refusal

**Bug Description** Real-time cyber safeguard false-positive — hard-blocks legitimate defensive malware analysis for a CVP-verified user. I'm doing pure STATIC deobfuscation of a captured JS malware dropper (Order_<n>.js) for incident response: decoding the packer layers and replaying the sample's own AES routine in Node/Python to dump the next stage to a fi

importance 4@FelixCmmrsvclosedbugplatform:linuxarea:modelstaleOriginal evidence ↗
issue

Allow adjusting font size of the chat/conversation panel independently from the editor

Body: **Is your feature request related to a problem? Please describe.** The Claude Code VS Code extension's chat/conversation panel shares font size with the rest of the editor. I have `editor.fontSize` set to a comfortable size for reading/writing code, but the same size feels too small for reading conversation text in the Claude Code panel. Currently ther

importance 4@GeraltRclosedenhancementarea:a11yarea:ideplatform:vscodestaleOriginal evidence ↗
issue

[FEATURE] Allow opening the terminal in a brand-new (empty) Code tab session

### Problem In the Claude Desktop app → **Code** tab, the terminal toggle button only appears **after** the session has at least one chat message. On a **brand-new / empty session** (before sending anything), there is no way to open or show the terminal at all. So if I want to run a few commands myself before starting the conversation, I cannot — I have to s

importance 4@pandhupclosedenhancementplatform:macosarea:desktopstaleOriginal evidence ↗