Stats
Actions
Tags
From agent-notch
Use when configuring or debugging the Agent Notch plugin for macOS local agent approvals, AskUserQuestion prompts, Codex requestUserInput, or QoderWork question and approval hooks.
How this skill is triggered — by the user, by Claude, or both
Slash command
/agent-notch:agent-notchThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Agent Notch wraps the Vibe Notch Swift UI as an installable OpenPlugin bundle
Agent Notch wraps the Vibe Notch Swift UI as an installable OpenPlugin bundle for Codex, Claude Code, and QoderWork hooks.
The primary runtime is a Vibe-compatible SwiftPM app:
hooks/scripts/hitl-island-hook.mjs normalizes Claude, Codex, and QoderWork hook payloads.vibe-runtime/Sources/AgentNotchVibe vendors the upstream Vibe Notch UI, session store, chat/history views, and hook socket server.vibe-runtime/Package.swift builds the app with SwiftPM, so no Xcode project is required.AskUserQuestion and Codex app-server requestUserInput prompts are bridged into Vibe QuestionRequest socket events and answered from the Vibe session row/chat view.hooks/scripts/agent-notch-manager.mjs remains a forced compatibility path when AGENT_HITL_USE_LEGACY_MANAGER=1.hooks/scripts/display-island.mjs remains a last-resort fallback when neither runtime can start.Use it for:
PermissionRequest hook flows.PermissionRequest hook flows.item/tool/requestUserInput result encoding.PermissionRequest hook flows.UserPromptSubmit, PreToolUse, PostToolUse, and Stop events show processing, running-tool, and ready states without using normal tools as approval fallbacks.QuestionRequest prompts for typed or choice-based user input.The hook entrypoint is hooks/scripts/hitl-island-hook.mjs. It sends status and
blocking permission/question events to the Vibe socket by default, then writes
the client-specific JSON response to stdout after Vibe resolves the request.
npx claudepluginhub zichaojin/opentasks --plugin agent-notchProvides behavioral guidelines to reduce common LLM coding mistakes, focusing on simplicity, surgical changes, assumption surfacing, and verifiable success criteria.
Searches, retrieves, and installs Agent Skills from prompts.chat registry using MCP tools like search_skills and get_skill. Activates for finding skills, browsing catalogs, or extending Claude.
Creates, edits, and optimizes skills for Claude Code, including drafting, evaluating with test prompts, iterating on performance, and improving skill descriptions for better triggering accuracy.