联网情报收集:识别组件后必做CVE/搜索引擎/中文社区/GitHub PoC/资产引擎/即时情报/依赖扩展+受阻换路。Use when a framework/component/version is identified and must search before exploit.

Upgrade the local Natron MacPorts portfile overlay against upstream MacPorts, and build Natron into a macOS arm64 DMG. Use when asked to update/upgrade the Natron MacPorts ports, run check.sh, reintegrate upstream Portfile changes, resolve failing Natron patchfiles after a MacPorts sync, regenerate Portfile.patch files, or build Natron / produce a DMG via tools/jenkins/launchBuildMain.sh. Enforces that lang/libomp stays pinned.

Presentation creation, editing, and analysis. When Claude needs to work with presentations (.pptx files) for: (1) Creating new presentations, (2) Modifying or editing content, (3) Working with layouts, (4) Adding comments or speaker notes, or any other presentation tasks

Visual language rules for Engram surfaces. Trigger: Any dashboard styling, typography, spacing, or visual identity change.

Commit and branch naming standards for Engram contributors, enforced by GitHub rulesets. Trigger: Any commit creation, review, or branch cleanup.

Solve a user-specified web task code-as-action style by driving a local Playwright browser through one bash command at a time, saving screenshots and an action log into `final_runs/run_<id>/`, and visually verifying the result. Use when the user asks to automate a web task (search, filter, form-fill, multi-step flow, data extraction) and wants reusable scripts plus screenshot evidence rather than a one-shot answer.

Create user stories with Mike Cohn format and Gherkin acceptance criteria. Use when turning user needs into development-ready work with clear outcomes and testable conditions.

Create a user story map that lays out activities, steps, tasks, and release slices. Use when planning a workflow, backlog, or MVP around the user journey.

Look up SaaS finance metrics, formulas, and benchmarks fast. Use when you need a quick metric definition, formula, or benchmark during analysis.

Frame an epic as a testable hypothesis with target user, expected outcome, and validation method. Use when defining a major initiative before roadmap, discovery, or delivery planning.

Create a company research brief with executive quotes, product strategy, and org context. Use when preparing for interviews, competitive analysis, partnerships, or market-entry work.

Four-phase debugging methodology with root cause analysis. Use when investigating bugs, fixing test failures, or troubleshooting unexpected behavior. Emphasizes NO FIXES WITHOUT ROOT CAUSE FIRST.

Remote-control tmux sessions for interactive CLIs by sending keystrokes and scraping pane output. Use when: (1) monitoring Claude/Codex sessions running in tmux, (2) sending input to interactive terminal applications, (3) scraping output from long-running processes inside tmux, (4) navigating tmux panes/windows programmatically, or (5) checking on background work in existing sessions. NOT for: one-off shell commands (use exec_command), starting new background processes (use background_process), or non-tmux interactive processes.

Optimizes the performance of existing Liger Kernel Triton kernels. Profiles kernels, diagnoses bottlenecks (memory-bound vs compute-bound), generates multiple optimization variants with benchmarking, and applies the best variant while maintaining correctness. Supports GPU architecture-specific optimization (Ampere, Hopper, Blackwell). Use when a user asks to optimize, speed up, tune, profile, or reduce memory of an existing Liger kernel.

Develops production-ready Triton kernels for Liger Kernel. Creates new kernels from PyTorch operations (local files, URLs, code snippets, or natural language) with ops, module wrappers, functional APIs, unit tests, benchmarks, and plots. Also modifies existing Liger kernels. Use when adding a new Triton kernel, converting a PyTorch operation to Triton, or updating an existing Liger kernel.

Reading companion agent. Accompanies user through any text (books, articles, essays, papers, news) with translation, structural annotation, deep questioning, and cross-domain insights. Detects language, translates English to Chinese (faithfulness-expressiveness-elegance), guides reader to understand the author and encounter real questions. Use when user says '伴读', '陪我读', '读这篇', 'read with me', 'companion read', or shares a text/URL wanting guided reading.

给一个领域，找出背后真正撑着它的几根独立的力。十几个现象砍到不可再少的生成器——砍完能把现象一个个生回来，才算数。Use when user says '降秩', '找秩', '秩是什么', '这个领域靠什么撑着', '背后是什么', or wants to decompose any domain to its irreducible generators.

极简演讲铸造器（Outline-Faithful）。把 orgmode/markdown outline 1:1 铸成单文件离线 HTML；以一页一语义动作、六种固定构图角色、留白预算和投影大字门槛控制密度，支持 black/red/yellow、hacker、hacker-dark、表格、ASCII、LaTeX、自适应与翻页笔。USE WHEN 用户要求讲这个、present、做成演讲、slides、标语流、宣言体、slogan、manifesto、按 outline 美化。NOT FOR 内容提炼、改写或企业 PPT。

给一个领域、专业、角色、产品或争论找出真正框住它的几条约束，判明它们属于世界/规则/解释（硬/软/自设），看这组约束如何定义身份、补全问题、框出解空间并解释实际行为；尤其用于区分目标相同但约束不同导致的方案分歧，识别被误当硬事实的旧解释。USE WHEN 用户说 '约束', '找约束', '这个领域的约束是什么', '本质约束', '什么框住了它', '硬约束还是自设约束', '边界在哪', '这行为什么这么做', 'constraint', '/ljg-constraint', or gives any domain/profession/role/product/debate and wants its defining constraints and solution space mapped. NOT FOR 往下挖生成器（用 ljg-rank，找的是生成力不是边界）、纵向深钻一个观点到底（用 ljg-think）、拆一本书（用 ljg-book）、一个概念解剖（用 ljg-learn）。

Use when implementing, fixing, refactoring, or otherwise modifying OpenChamber source code, dependencies, exports, build configuration, generated assets, package contracts, or module ownership.