code-yeongyu

MUST USE for frontend/web UI/UX/visual work: building, styling, redesigning pages/components, React setup, performance audits, visual QA, taste, and polish. Routes four rulesets: design taste router and brand references; perfection for Playwright/Chromium Lighthouse/Core Web Vitals; ui-ux-db palettes/fonts/guidelines; designpowers personas/accessibility/critique/handoff; plus curl-only lazyweb real-app-screen research for design direction. Triggers: frontend, UI, UX, design, redesign, styling, layout, animation, motion, premium, luxury, minimal, brutalist, Awwwards, DESIGN.md, mockup, React, Lighthouse, accessibility, WCAG, Core Web Vitals, looks generic, make it pretty, like X brand, lazyweb, design research.

MUST USE for frontend/web UI/UX/visual work: building, styling, redesigning pages/components, React setup, performance audits, visual QA, taste, and polish. Routes four rulesets: design taste router and brand references; perfection for Playwright/Chromium Lighthouse/Core Web Vitals; ui-ux-db palettes/fonts/guidelines; designpowers personas/accessibility/critique/handoff; plus curl-only lazyweb real-app-screen research for design direction. Triggers: frontend, UI, UX, design, redesign, styling, layout, animation, motion, premium, luxury, minimal, brutalist, Awwwards, DESIGN.md, mockup, React, Lighthouse, accessibility, WCAG, Core Web Vitals, looks generic, make it pretty, like X brand, lazyweb, design research.

Persistent Codex-style goal tracking for pi. Use when the user explicitly asks to set, continue, audit, pause, resume, complete, or inspect a long-running goal.

MUST USE for planning before coding: 5+ steps, ambiguous scope, multiple modules, architecture decisions, a vague 'just make it good / figure out what to build' brief, or any request to plan, interview, or break work down. Explore-first planning consultant (Prometheus) that grounds in the codebase, asks only the forks exploration cannot resolve - or researches them to best practice when the intent is fuzzy - waits for explicit approval, then writes ONE decision-complete work plan a worker executes with zero further interview. Triggers: ulw-plan, plan this, make a plan, plan before coding, interview me, break this down, start planning, plan mode, just make it good, figure out what to build.

MUST USE for planning before coding: 5+ steps, ambiguous scope, multiple modules, architecture decisions, a vague 'just make it good / figure out what to build' brief, or any request to plan, interview, or break work down. Explore-first planning consultant (Prometheus) that grounds in the codebase, asks only the forks exploration cannot resolve - or researches them to best practice when the intent is fuzzy - waits for explicit approval, then writes ONE decision-complete work plan a worker executes with zero further interview. Triggers: ulw-plan, plan this, make a plan, plan before coding, interview me, break this down, start planning, plan mode, just make it good, figure out what to build.

MUST USE for any real runtime debugging across ANY language or binary — crashes, silent failures, wrong responses, stuck processes, memory leaks, async misbehavior, unexplained timing, reverse engineering. Runs a hypothesis-driven loop: form ≥3 hypotheses, investigate in parallel, after 2 failed rounds spawn Oracles from orthogonal angles, confirm root cause, lock with a failing test, fix minimally, QA by actually USING the system, scrub artifacts. The actual HOW lives in `references/` — READ THEM. Triggers: 'debug this', 'why is X not working', 'hanging', 'attach a debugger', 'reverse engineer', 'pwndbg', 'gdb', 'lldb', 'node inspect', 'tsx debug', 'pdb', 'dlv', 'delve', 'rust-gdb', 'set a breakpoint', 'context window exploded', 'why is the response empty', 'attach the debugger', 'debug it', 'why is this happening', 'trace this bug', 'reproduce and fix', 'silent failure', 'HTTP 200 but empty', 'why did it stop', 'inspect the binary', 'reverse engineering', 'playwright'.

MUST USE for ANY work on .py .pyi .rs .ts .tsx .mts .cts .go files. One philosophy: strict types, modern stacks (Pydantic v2 / serde+thiserror / Zod / gin+sqlc+pgx+slog), modern toolchains (uv+basedpyright+ruff / cargo+clippy+miri / Bun+Biome+tsc / gofumpt+golangci-lint v2+nilaway+go-race), parse-don't-validate, exhaustive match, typed errors, no any/unwrap/panic, 250 LOC ceiling, TDD, consumer-routed logging. Routes to references/{python,rust,typescript,rust-ub,go}/ + references/logging.md. Triggers: write/edit Python/Rust/TypeScript/Go code, new project, gin server, bubbletea TUI, CJK IME, connect-go RPC, sqlc pgx, branded ids, exhaustive match, unsafe Rust, miri, oversized file, refactor, TDD, e2e test, logging, log levels, structured logging, observability, arena, allocator, bumpalo, const fn, const generics, comptime, zero-alloc, bitfield, repr, scopeguard, errdefer, Zig-like, zerocopy, packed struct.

Binding ultrawork mode directive for omo on Codex. When a prompt contains ultrawork or ulw, the omo UserPromptSubmit hook injects a short bootstrap that points at this file. Read the whole file and follow every rule in it for the rest of the task.

MUST USE for frontend/web UI/UX/visual work: building, styling, redesigning pages/components, React setup, performance audits, visual QA, taste, and polish. Routes four rulesets: design taste router and brand references; perfection for Playwright/Chromium Lighthouse/Core Web Vitals; ui-ux-db palettes/fonts/guidelines; designpowers personas/accessibility/critique/handoff. Triggers: frontend, UI, UX, design, redesign, styling, layout, animation, motion, premium, luxury, minimal, brutalist, Awwwards, DESIGN.md, mockup, React, Lighthouse, accessibility, WCAG, Core Web Vitals, looks generic, make it pretty, like X brand.

Maximum-saturation research orchestration: parallel explore+librarian swarms across codebase, web, official docs, and OSS repos; a recursive EXPAND loop driven by leads workers return in message text; empirical verification by running code; cited synthesis and optional MD/HTML/PDF/PPTX reports. ACTIVATES ONLY on an explicit user demand for research — the word 'ulw-research' ('/ulw-research', '$ulw-research'), any 'ulw' research wording, or an explicit request for research / deep research / an ultra-precise investigation, in any language. Never self-activates for ordinary questions, debugging, or implementation context-gathering. While active it overrides exploration-bounding defaults: exhaustive coverage is the goal.

MUST USE for ANY work on .py .pyi .rs .ts .tsx .mts .cts .go files. One philosophy: strict types, modern stacks (Pydantic v2 / serde+thiserror / Zod / gin+sqlc+pgx+slog), modern toolchains (uv+basedpyright+ruff / cargo+clippy+miri / Bun+Biome+tsc / gofumpt+golangci-lint v2+nilaway+go-race), parse-don't-validate, exhaustive match, typed errors, no any/unwrap/panic, 250 LOC ceiling, TDD. Routes to references/{python,rust,typescript,rust-ub,go}/. Triggers: write/edit Python/Rust/TypeScript/Go code, new project, gin server, bubbletea TUI, CJK IME, connect-go RPC, sqlc pgx, branded ids, exhaustive match, unsafe Rust, miri, oversized file, refactor, TDD, e2e test, arena, allocator, bumpalo, const fn, const generics, comptime, zero-alloc, bitfield, repr, scopeguard, errdefer, Zig-like, zerocopy, packed struct.

Remove AI-generated code smells (slop) from branch changes or an explicit file list. Locks behavior with regression tests FIRST, then runs categorized cleanup via parallel `deep` agents in batches of 5, then verifies with quality gates. Covers 10 slop categories including performance equivalences, excessive complexity (object annotations, if/elif variant chains), and oversized modules (250+ pure LOC with mandatory modular refactoring). MUST USE when the user asks to "remove slop", "clean AI code", "deslop", "clean up AI-generated code", "remove AI slop", or wants to clean up AI-generated patterns from recent changes. Triggers - "remove ai slops", "clean ai code", "deslop", "cleanup AI generated", "remove AI slop", "clean up AI-generated code", "strip slop", "ai-slop cleanup".

MUST USE after building/changing any UI or when asked whether a page, component, or TUI looks right. Rigorous visual QA across web/page and terminal UIs. Prefer browser:control-in-app-browser for unauthenticated browser/page QA in Codex, then Playwright/agent-browser/dev-browser. Captures screenshot/TUI evidence with bundled diff scripts, runs design-system/functional and visual-fidelity/CJK reviewer passes, then synthesizes a good/bad verdict. Triggers: visual QA, screenshot/pixel diff, UI looks wrong, reference fidelity, design system check, responsive check, CJK text clipping, TUI alignment, box-drawing drift.

MUST USE for planning before coding: 5+ steps, ambiguous scope, multiple modules, architecture decisions, a vague 'just make it good / figure out what to build' brief, or any request to plan, interview, or break work down. Explore-first planning consultant (Prometheus) that grounds in the codebase, asks only the forks exploration cannot resolve - or researches them to best practice when the intent is fuzzy - waits for explicit approval, then writes ONE decision-complete work plan a worker executes with zero further interview. Triggers: ulw-plan, plan this, make a plan, plan before coding, interview me, break this down, start planning, plan mode, just make it good, figure out what to build.

(builtin) Initialize hierarchical AGENTS.md knowledge base

Intelligent refactor command. Triggers: refactor, refactoring, cleanup, restructure, extract, simplify, modernize.

Use when the user asks about Codex Rules behavior, injected project rules, supported rule file locations, matching, or environment configuration.

Goal-like loop that uses ultrawork mode to decompose work into systematic, evidence-bound steps.

Removes AI-generated code smells from a SINGLE file while preserving functionality. For multiple files, call in PARALLEL per file.

MUST USE for any real runtime debugging across ANY language or binary — crashes, silent failures, wrong responses, stuck processes, memory leaks, async misbehavior, unexplained timing, reverse engineering. Runs a hypothesis-driven loop: form ≥3 hypotheses, investigate in parallel, after 2 failed rounds spawn Oracles from orthogonal angles, confirm root cause, lock with a failing test, fix minimally, QA by actually USING the system, scrub artifacts. The actual HOW lives in `references/` — READ THEM. Triggers: 'debug this', 'why is X not working', 'hanging', 'attach a debugger', 'reverse engineer', 'pwndbg', 'gdb', 'lldb', 'node inspect', 'tsx debug', 'pdb', 'dlv', 'delve', 'rust-gdb', 'set a breakpoint', 'context window exploded', 'why is the response empty', 'attach the debugger', 'debug it', 'why is this happening', 'trace this bug', 'reproduce and fix', 'silent failure', 'HTTP 200 but empty', 'why did it stop', 'inspect the binary', 'reverse engineering', 'playwright'.

Designer-turned-developer who crafts stunning UI/UX even without design mockups

(builtin) Initialize hierarchical AGENTS.md knowledge base

Strategic planning consultant that produces decision-complete work plans through Socratic interview, codebase exploration, Metis gap analysis, and optional Momus high-accuracy review. MUST USE when the task has 5+ steps, scope is ambiguous, multiple modules are involved, or the user asks for a plan. Triggers: plan this, create a work plan, interview me, start planning, prometheustic, plan mode, help me plan this, break this down.

MUST USE for ANY work on .py .pyi .rs .ts .tsx .mts .cts .go files. One philosophy: strict types, modern stacks (Pydantic v2 / serde+thiserror / Zod / gin+sqlc+pgx+slog), modern toolchains (uv+basedpyright+ruff / cargo+clippy+miri / Bun+Biome+tsc / gofumpt+golangci-lint v2+nilaway+go-race), parse-don't-validate, exhaustive match, typed errors, no any/unwrap/panic, 250 LOC ceiling, TDD. Routes to references/{python,rust,typescript,rust-ub,go}/. Triggers: write/edit Python/Rust/TypeScript/Go code, new project, gin server, bubbletea TUI, CJK IME, connect-go RPC, sqlc pgx, branded ids, exhaustive match, unsafe Rust, miri, oversized file, refactor, TDD, e2e test, arena, allocator, bumpalo, const fn, const generics, comptime, zero-alloc, bitfield, repr, scopeguard, errdefer, Zig-like, zerocopy, packed struct.

Create YAML format work plans saved as .sisyphus/tasks/{name}.yaml with strict schema validation. Analyze user requirements, gather project context, and generate structured plans with verification specs. ALWAYS includes mandatory plan-reviewer verification. Use when users request YAML-based work planning or Sisyphus-compatible task breakdown.

Activate when user expresses intent to start Sisyphus work (e.g., "okay then work like that", "let's execute this", "start working on this"). Extract ai-todolist.md content from GitHub comments/files, create branch, commit, and trigger GitHub Actions workflow. If no ai-todolist format found, load plan-writer skill instead.

Patterns for controlling interactive debuggers (pdb, ipdb, gdb, lldb, node debug) via terminalcp

Patterns for managing and monitoring long-running processes (builds, tests, servers, etc.) via terminalcp

Patterns for controlling REPL sessions (Python, IPython, Node.js, Ruby, etc.) via terminalcp

Patterns for controlling TUI applications (vim, htop, tig, lazygit, etc.) via terminalcp

Common Playwright automation patterns for web testing and scraping via MCP

Web testing patterns and best practices using Playwright for automated testing

Manual QA harness for the senpi coding agent itself. MUST USE after changing packages/ai, packages/agent, packages/coding-agent, or packages/tui — a green typecheck and `npm test` are NOT QA. Drives the real CLI from source in an isolated sandbox (never touches ~/.senpi or real credentials) across four channels: remote RPC (--mode rpc JSONL stdio), TUI smoke (node-pty on Windows, tmux on POSIX), mock loop (a local fake model server for deterministic, zero-token agent-loop runs), and CLI smoke (--help/--print/--list-models). Every helper ships a --self-test. Use whenever someone says qa senpi, test the agent, verify my change, rpc qa, tui qa, mock-loop qa, smoke the cli, or needs evidence that an agent-loop, tool, keybinding, or provider change works end to end. Capture evidence to local-ignore/qa-evidence/.

Sync a fork branch with an upstream remote using a history-preserving merge. Use this whenever the user says /merge-upstream, merge upstream, sync upstream, sync fork, or wants upstream changes integrated without rebasing or force-pushing.

Implements a single feature in the pi-mono `todotools` builtin extension work. Use for refactoring, continuation runtime, config resolver, prompt builder, test authoring, golden snapshots, CHANGELOG entries, and harness helpers. Does NOT do manual tmux QA.

Run a single manual tmux-based QA scenario for the todo continuation feature against the real CLI (./pi-test.sh) in an interactive TUI. Captures scrollback, asserts deterministic pass/fail count markers, and cleans up test fixtures. Use only for the manual-qa milestone features.

Example skill loaded from resources_discover

Drive the user's host machine directly through Cua, with no sandbox. Maps to Cua's Localhost API. Higher risk - the agent can click, type, and run shell commands on the real desktop. Use only when the user explicitly opts in and read the safety section first. Triggers - "control my computer", "click here", "use my real Chrome", "no sandbox", "내 컴퓨터 자동화".

Mouse, keyboard, and scroll primitives for cua_click / cua_type / cua_key / cua_scroll. Reference for action shapes, modifier keys, key chord syntax, and screenshot-driven targeting workflows.

pi-cua-integration is a Pi/Senpi extension that bridges Cua (trycua/cua) for computer-use sandboxes and host control. MUST USE when the user asks to drive a desktop UI, take screenshots, run a "computer use" task, control a sandbox, or invoke any cua_* tool. Triggers - "cua", "computer use", "sandbox", "screenshot the desktop", "click the screen", "take over the browser", "샌드박스 켜줘", "스크린샷 찍어", "내 컴퓨터에서 자동화".

Use Cua's cloud sandboxes via cua.ai. Requires CUA_API_KEY. Cloud sandboxes are isolated VMs run by trycua with billing per usage. Use when the user wants no local Docker/QEMU, wants Windows/macOS guests without Apple Silicon, or explicitly asks for cloud.

Run Cua sandboxes locally - no API key required. Covers Docker (XFCE/Kasm) for Linux containers, QEMU for full VMs, Lume for macOS guests on Apple Silicon, and Tart for Apple VZ VMs. Use when the user wants a local sandbox, wants to keep things off the cloud, or asks "spin up an Ubuntu desktop", "open a sandbox VM", "let me try this safely first".

Commits changes in atomic units following dependency order. Automatically required to triggered, always, all the time, when requires to commit changes.

Specialized GitHub PR intelligence agent for automatically gathering comprehensive context from Pull Requests. Activate when users need CI failure analysis, review comment investigation, or PR status assessment. Triggers on requests like "CI 실패 원인 찾아줘", "gather review comments", "check PR status", "analyze PR".

GitHub Pull Request creation specialist. Analyzes user requirements to create PRs with structured titles and bodies matching the user's query language. Handles git change analysis, PR draft creation, user confirmation, and final PR creation via gh CLI.

Convert and share Claude Code conversation sessions as readable markdown files. Use when user wants to share a session transcript, export conversation history, or create a shareable markdown document from a Claude Code session. Triggered by requests like "share this session", "export conversation", "convert session to markdown".

GitHub Pull Request creation specialist. Analyzes user requirements to create PRs with structured titles and bodies matching the user's query language. Handles git change analysis, PR draft creation, user confirmation, and final PR creation via gh CLI.

Use for macos-cua setup, troubleshooting, MCP/CLI operation, or desktop automation only when Senpi/pi-extension registered tools are unavailable. In Senpi registered-tool mode, do not load this skill just to operate the UI; use the tool guides on `get_app_state`, `click`, `type_text`, `press_keys`, `scroll`, `drag`, `set_value`, `select_text`, `perform_secondary_action`, and `zoom`. Triggers: macos-cua setup, macos-cua troubleshooting, macos-cua CLI, macos-cua MCP, macos computer use without registered tools, host-native computer use, ScreenCaptureKit, CGEvent, macOS GUI automation fallback.

Persistent Codex-style goal tracking for pi. Use when the user explicitly asks to set, continue, audit, pause, resume, complete, or inspect a long-running goal.

Manage Akiflow tasks from CLI. Use when creating, listing, completing tasks, managing projects, or viewing calendar/time blocks. Credential extraction from browser—no API keys needed.

MUST USE when the user needs current, online, or web-derived information — news, latest releases, vendor docs, real-world examples, recent changes, or anything not in the model's training set. LLM-neutral skill that exposes a unified web search CLI behind a single command. Runs single or multi-provider in parallel, saves raw + normalized JSON to a temp directory, prints the path so callers can pipe through rg / jq / grep. Zero-config out of the box. Works identically on macOS, Linux, WSL, Git Bash for Windows, and any environment with Python 3.8+. Triggers: 'search the web', 'look up online', 'google this', 'find current information', 'latest news', 'recent docs', 'check the internet', 'web search', 'browser search', 'research this', 'what is the latest', 'who is X', 'find examples of', 'is there documentation for', 'how do people solve', 'web-search', 'web_search'.