Hooks & automation — page 2
Issues 201–400 of the 9,356 matching this topic. The trend chart and topic overview are on page 1.
- Feature Request: Add Hook for Tool Execution Failures (`OnToolError` / `ToolUseFailure`)
- [BUG] Plugins not loading
- Support ".claudeignore" file to prevent secret exposure
- [Feature Request] Support inline model effort level specification in prompts
- [FEATURE] UserInputChange hook — expose the prompt input buffer to hooks, mirroring MessageDisplay
- [Feature] Reconnect/restart MCP servers from Claude Code in Claude Desktop without Cmd+Q (expose /mcp UI or auto-reconnect)
- [FEATURE] Allow bypassPermissions mode to fully bypass .claude/ directory protections (or provide an opt-in override)
- [BUG] --dangerously-skip-permissions does not bypass "modify config files" prompt for ~/.claude/ writes
- Auto-compact fails with Opus 4.5 extended thinking blocks
- [BUG] Account-level plugins never sync to Claude Code on the web (plugins_sync count: 0) while account skills sync fine
- [FEATURE] Treat Plugins as First-class Vendored Dependencies
- Feature request: bring back /buddy as a permanent feature
- [FEATURE] Expose pasted image data to hooks and plugins
- [FEATURE] /pr_comments ranks higher than /commit when typing /com - improve fuzzy matching, enter selects unexpected command and allow disabling built-in commands
- Make autonomous Claude Code actually viable: tiered Opus brains + Sonnet workers + persistent state
- [FEATURE] API / sync mechanism for Claude Team Project knowledge files
- [FEATURE] PermissionRequest and PermissionResponse hook events for metrics
- Feature Request: Add PreCompact and PostCompact hooks for custom context management
- [Bug] VSCode Extension: User-level rules (~/.claude/rules/) not loading into context
- Remote Control regression: server-injected `<system-reminder>Message sent at …>` prefix breaks slash commands from mobile/web
- [BUG] Cowork /schedule skill fails — create_scheduled_task tool not injected into session context (Windows)
- [BUG] updatedInput PreToolUse response does not work when multiple PreToolUse hooks are executed
- Feature Request: Expose usage quota in status line JSON
- [BUG] Custom slash commands can't request the user's permission to execute bash scripts
- [FEATURE] Date/Time Injection into prompts
- Claude Code suggests non-existent "fan out subagents" command
- VS Code webview shows `Unhandled case: [object Object]` and stops rendering while CLI backend keeps running
- Add opt-out for forced-worktree enforcement on background-session Edit/Write tools
- [Bug] EnterWorktree tool ignores WorktreeCreate/WorktreeRemove hooks
- [FEATURE] Add Skill Autocomplete to CLI (Parity with VS Code Extension)
- [FEATURE] Plugins - Add postInstall, postUpdate and postUninstall hooks in plugin.json
- New Hook suggestion: PostCompact
- Task tool fails with "Last message was not an assistant message" error when Stop hook is configured
- /code-review skill: silent fallback to main...HEAD reviews other people's commits, and JSON-only output is hard to read
- [FEATURE] /new slash command to switch to a fresh session while preserving the current one
- [BUG] Bash tool fails with zsh + zoxide configuration - "__zoxide_z not found" errors
- Let the Code tab's usage ring report session context, not only the 5-hour window
- [FEATURE] PermissionRequest hook: allow setting custom description when returning permissionDecision: "ask"
- [Critical] Max Plan subscriber: Opus 4.6 instruction-following regression breaks production workflows — 24-hook enforcement system cannot compensate for model-level degradation
- [Bug] v2.1.30 regression: CLI freezes when SessionStart hook errors (Windows)
- Add subscription usage data to statusline context
- [FEATURE] CLI support for .worktreeinclude file (parity with desktop app)
- Claude Code doesn't respond to MCP notifications/prompts/list_changed for dynamic command updates
- # [BUG] Slash command namespacing with subdirectories not working as documented
- [FEATURE] Add /fork (conversation branching) support to VS Code extension
- [BUG] Windows Terminal input completely frozen starting from v2.1.30 (regression from v2.1.29)
- Feature Request: Add Hooks for Git Workflow Automation (`PreCommit`, `PostCommit`)
- [BUG] Autocomplete UI view not working with any slash commands
- [BUG] Remote Control returns "disabled by your organization's policy" on Team plan despite admin toggle being ON
- feat: terminal bell (BEL) when Claude Code is waiting for tool approval
- [FEATURE] Expose Claude.ai plan usage in statusLine JSON payload
- [BUG] Commands are now apparently "skills" and are loaded into context without being invoked. This effectively removes command functionality from Claude Code
- [BUG] @ command includes .git files in autocomplete
- Feature Request: Run Claude with a slash command from CLI and provide arguments
- [BUG] PreToolUse permissionDecision: "allow" no longer suppresses prompt for Bash with dangerouslyDisableSandbox: true (2.1.116+ regression)
- [BUG] IS_DEMO env var suppresses workspace trust prompt without granting trust, breaking statusline and hooks
- [Bug] /compact fails with "Tool names must be unique" even after disabling all MCP servers (v2.0.70)
- [BUG] IMPORTANT: 2.0.30 please KEEP the output-style
- [FEATURE] Add deep research mode as a /research slash command
- [FEATURE] Pre-Exit Hook ( or bring us back Automatic Cost Display on Exit)
- AskUserQuestion UI not rendered when called after Skill tool invocation
- Claude Code hangs indefinitely on startup in non-git directories (regression since 2.1.31)
- [BUG] SessionStart hooks break stdin on Windows Terminal
- v2.1.23+ causes 100% CPU hang with multiple parallel instances and hooks
- Feature request: /cd command to change working directory mid-session
- [BUG] Grep tool causes "hook execution cancelled" errors that make Claude Code unusable
- [FEATURE] Make Workflow scripts (.claude/workflows/*.js) a distributable plugin component
- [BUG] Stop hook causes empty result in print mode (-p)
- [FEATURE] OAuth tokens should be revoked server-side on claude logout and on devcontainer shutdown to support secure workflows
- CLI: permissionDecisionReason not displayed in permission prompt for PreToolUse 'ask' decisions
- Slash commands do not compose
- [BUG] VS Code extension never draws mod UI (Pane, AbovePrompt, status, toast) although the engine accepts the vscode surface
- Write/Edit tools fail with EEXIST on OneDrive directories (v2.1.69 regression)
- [BUG] Claude Desktop "Something went wrong loading this session" - React #185 infinite loop on specific project directory
- LSP plugins not registering servers - LSP Manager initializes before plugins load
- [BUG] /context executes /compact instead, unless I ask nicely :-)
- Hook executor doesn't set CLAUDE_PLUGIN_ROOT environment variable
- Support a "Start" hook
- Background worker crash-loop on session start: "worker crashed (exit 1 — exit_with_message) — respawning…"
- [BUG] Tool-call results empty in live UI then flush late/out-of-order (Bash, Read, MCP, advisor) — macOS, Opus 4.8, parallel batches
- Agent View: home page input frozen after returning from agent detail (Windows)
- Channel notifications dropped during hook execution
- [BUG] Slash commands with Skill() in allowed-tools trigger rate limit on minimal diffs (regression)
- Remote Control not enabled for Max 20x subscriber ($200/mo)
- [BUG] SubagentStart/SubagentStop unreliable for settings.json hooks — cascade failures break agent lifecycle management
- [BUG] Slash commands not working - "Unknown skill" error for all built-in commands
- [BUG] Plugin hooks registered but not firing in VSCode extension
- Project scope MCP servers not displayed in claude mcp list command
- [BUG] Worktree isolation hard-blocks ALL `git -C <main-checkout>` calls — even read-only, even after a PreToolUse hook explicitly approves it
- Max plan: skills with model frontmatter blocked by 1M context extra-usage gate
- [BUG] CLAUDE.md Instructions Systematically Ignored / Suspected Model Substitution - Opus 4.6
- [BUG] deniedMcpServers serverName regex rejects plugin MCP server names
- [FEATURE] Plan Lifecycle Hooks
- Support slash command autocomplete when '/' is typed mid-prompt
- [Bug] Stray token "count" injected before tool calls causes parse failure in Claude Code
- [BUG] Plugin skills not appearing in `/` menu after v2.1.110 update — only `schedule` shows
- VS Code extension skill validator rejects valid frontmatter fields
- Remote Control unavailable on Max plan (v2.1.59)
- UserPromptSubmit agent hooks fail: "Messages are required for agent hooks"
- Anthropic API: Unexpected Reduction in Claude Code MAX Plan Usage Limits
- AskUserQuestion auto-selects default answer after ~60s idle — no way to configure or disable (v2.1.198)
- Agent Teams: lead session stops processing teammate responses until manual keystroke (interactive TTY, default mode)
- /loop command returns 'Unknown skill: loop' in CLI v2.1.71
- Feature Request: Programmatic API for enabling/disabling MCPs without restart
- Consider merging Skills and Slash Commands
- [BUG] A complete CLAUDE.md rule contract governed nothing: 9 fabricated causes, stale state asserted as current, acceptance step silently skipped across a 4.5h session
- [BUG] /goal Stop hook fails with "JSON validation failed" when prompt-based hook returns markdown-fenced JSON; goal never auto-clears
- Post-mortem 2026-04-28: 12 multi-agent coordination bugs surfaced across a single autonomous-overnight cycle
- [FEATURE] CLI task list (Ctrl+T) should allow viewing all tasks, not just the top 5
- bypassPermissions mode still prompts for Edit/Write on UNC paths (Windows)
- [BUG] Project settings hooks not loaded when working directory is a subdirectory of the repo
- [BUG] /context command output renders twice
- [BUG] Slash commands only load first file on ExFAT filesystems due to JavaScript Number precision loss
- [Bug] Hookify plugin: incorrect sys.path configuration breaks module imports
- [FEATURE] Autocomplete Hook or Per-Project File Include
- [Feature Request] Add automatic cost reporting on session exit
- [BUG] Hook Regression in 2.0.17-2.0.22 - UI Shows Duplicate Messages for Stop, UserPromptSubmit, SessionStart (worked in 2.0.15)
- [BUG] Read(**/.env) deny rule prompts on every rg directory search, though rg skips hidden files by default
- Assistant text before/between tool calls is not rendered in the CLI, while hooks and the transcript still see it
- [FEATURE] iOS: start new Claude Code sessions on my paired Mac, with full feature parity
- [FEATURE] Raise, remove, or make configurable the 10,000-char limit that persistHookOutput enforces on hook additionalContext (introduced in v2.1.89)
- PreToolUse/PostToolUse hooks do not fire for subagent (Agent tool) tool calls
- [BUG] Worktree cleanup prompt isn't used when custom worktree hooks are enabled
- [FEATURE] Suppress async hook completion message for Stop/SubagentStop events
- Custom agent definitions (.claude/agents/) are silently ignored for team agents
- MCP OAuth tokens not auto-refreshing despite valid refresh tokens
- Windows: plugin hooks spawn visible cmd.exe/conhost.exe windows despite windowsHide:true
- [Bug] API Error 400 - Tool Use Concurrency Issues - Widespread Regression Affecting All Platforms
- # [BUG] Extreme memory and CPU usage on WSL2 causing system instability
- [DOCS] Clarify relationship between skills and slash commands post v2.1.1
- [BUG] claude code cli hooks not called in 1.0.98
- Critical Security Bug: `deny` permissions in `settings.json` are not enforced**
- Typed input queued mid-turn is silently dropped at turn end (end_turn, no Escape involved) -- interactive TUI
- [BUG] Claude Code in the Claude Desktop app runs UserPromptSubmit hooks and enforces blocks, but renders no user-facing hook feedback (CLI renders all)
- 2.1.141: Notification:permission_prompt hook event stops firing during active thinking (regression from 2.1.139)
- VS Code: Edit tool prompts for permission despite bypassPermissions mode
- [BUG] Claude stops responding mid-task after successful tool calls
- [BUG] [1M] Opus 4.6 1M context unavailable on Max 5x — works on another machine with same account
- Feature Request: Layered memory system for persistent cross-session context
- [BUG] Sandbox `denyWithinAllow` list omits `.git/` prefix from git-internal file paths
- [BUG] Edit/Write tools bypass permissions.ask rules (regression of #11226)
- Link to JSON Schema for settings.json in the official docs
- [BUG] Hooks not loading from settings.json - /hooks shows "No hooks configured yet" despite valid configuration
- Subagent launch fails with 'Tool names must be unique' when MCP server is configured
- [BUG] "Stop hook error" displayed despite hooks producing zero output (v2.0.28)
- [BUG] Claude Code TS SDK expects the prompt stream to remain open while working
- Hot reload of agents and slash commands
- Model Selection Fails for Slash Commands with Anthropic API
- Feature Request: Enhance Hooks with Integrated Types (MCP Tool, Inline Script, API Call)
- [BUG] Slash Commands not recognized
- Slash Command Autocomplete Fails for @ File References
- [FEATURE] Keyboard shortcut for microphone in main chat window (Claude Desktop macOS)
- [FEATURE] Enable voice dictation (/voice) when using AWS Bedrock, Vertex AI, and other third-party providers
- [Feature] Multi-agent runtime needs mechanical enforcement: 9 gaps that defeat unattended overnight operation
- [desktop] Add /reload-plugins command for in-process plugin refresh (parity with CLI)
- [BUG] False "Hook Error" labels cause Claude to prematurely end turns (exit 0, no stderr, valid JSON)
- Claude Code destroyed user's uncommitted work by running git reset --hard on session startup — TWICE
- [BUG] Hooks fail with posix_spawn '/bin/sh' ENOENT due to sandbox restrictions
- [BUG] Installing plugins on Claude Code for Web Envs not working as expected
- Feature Request: PreSessionEnd Hook
- [BUG] Running File Read on Windows is really slow, almost unusable. Please optimize the cygpath logic!
- [BUG] Skills System: Natural Language Intent Matching Completely Broken
- [BUG] Claude Code does not detect slash commands located under .claude/commands directory
- [BUG] `CallToolResult.structuredContent` is ignored
- [FEATURE] VS Code extension: draw function-hook mod UI (AbovePrompt band, Pane)
- [FEATURE] Routine edit/create form: expose effort level, not just model
- Sandbox filesystem protection leaves stray placeholder files / bind mounts in project working tree
- [BUG] Opus 4.8 confabulates user messages, a fake "prompt injection attack" narrative, and fabricated tool/host facts in long sessions (2 sessions, JSONL-verified)
- [FEATURE] Inverse of background mode: archive agent from Agent View without destroying data
- feat(plugins): namespace plugin skills with plugin name prefix like commands
- [FEATURE] Allow forcing extended thinking on Opus 4.7
- EnterWorktree tool creates worktrees inside .claude/ directory, breaking slash command and skill discovery
- permissions.allow and PreToolUse hooks cannot override .claude/ sensitive-file prompt (blocks automation)
- [BUG] /reload-plugins does not load skills from newly installed marketplace plugins
- [FEATURE] Make long commands easy copy-pasteable in terminal
- [FEATURE] Allow existing Claude Code sessions to join a team as teammates
- Windows: Cannot type in terminal until pressing Enter after v2.1.31 update (regression)
- Sandbox: zsh writes to /tmp/claude-501/cwd-* causing 'operation not permitted' errors
- [FEATURE] Add setting to disable automatic background task completion notifications
- Feature request: CLI flags/settings to hide default slash commands and agents from menu
- [BUG] ralph-wiggum stop hook triggered in separate session
- [BUG] undefined is not a constructor (evaluating 'new G($,I,L)')
- [BUG] MCP servers from marketplace no longer available
- [FEATURE] A way to switch models mid-session that does not persist to the settings
- [FEATURE] Add tool alias configuration for modern CLI alternatives
- [BUG] Prompt-based Stop hook cannot response correct JSON
- [BUG] Claude Code not automatically loading .claude/settings.json during startup despite documentation indicating this should happen
- Auto mode: let a classifier block fall back to a permission prompt instead of a hard deny
- disable-model-invocation: true breaks explicit /skill-name invocation
- [FEATURE] Let the agent trigger /compact itself (agent-invokable compaction)
- Remote Control: slash commands (/clear, /compact, /context, /bug) route to model as plain text instead of executing
- /desktop command fails on Windows Store (MSIX) installation
- Skill tool call returns 'Tool result missing due to internal error' and agent gets stuck
- attribution setting does not control session URL in commit messages
- Feature: Allow whitelisting sites (including localhost) in Claude in Chrome extension
- Move behavioral instructions out of tool definitions to reduce fixed token overhead
- cleanupPeriodDays: 0 silently disables all transcript persistence (docs say it disables cleanup)
- [FEATURE] CLAUDE.md instructions are read but not reliably followed - need enforcement mechanism
- [FEATURE] Native Git Worktree Management for Parallel & Isolated Agent Execution
- ralph-loop plugin: Windows compatibility issue - stop-hook.sh cannot execute