Skills directory

Serves an LLM on a supported AMD EPYC server CPU using vLLM with zentorch, in Docker, Podman, or conda. Use for "vLLM on CPU", "zentorch serving", or an EPYC CPU endpoint, including on a host that also has AMD Instinct GPUs. Detects the EPY

Serves AI models on AMD Instinct GPU hardware using vLLM. Use this skill whenever the user wants to run, serve, deploy, start, host, or launch a language model on an AMD GPU, AMD Instinct, MI300X, MI325X, MI350X, or MI355X. Also use when th

Use when lasting learnings from the current session should be recorded in agent instruction files - AGENTS.md, CLAUDE.md, or the user's global instructions - such as pitfalls hit, mistakes corrected, conventions found, or commands that fina

masseater/typescript-template から作ったリポジトリを、テンプレートに揃える。GitHub 側の設定を gh で揃え、中身の移行が完了しているかを、テンプレートの骨格が入り、効き、古い形が消えているかで判定する。新しいリポジトリを作ったとき、AI に移行を任せたあとに使う。ほかのリポジトリでも使えるが、その場合は適用する手順をユーザーに確認する。

Code commit, PR creation, merge, and issue closure workflow via GitHub CLI (gh). Triggers after a goal (GitHub Issue) implementation is complete — commit code, push branch, create PR, merge, then close the issue. Use when the user says "提交代

Agent Skills の skill、つまり `SKILL.md` とそのディレクトリを作る・直す・添削するときに読む。本文を skill が必要な理由と達成したい状態から導いて書き、直すときも全体を導き直す。

スキルの改良を正本リポジトリに提案する。スキルどおりに進まない、Web 検索などでスキルと違う情報や新しい情報が得られた、スキルの条件を厳守すると最良の結果が得られない、と判断したときの進め方を定義する。ホームに導入したスキルを直接編集しないこと、提案先の振り分け、Issue のタイトルと本文テンプレートを含む。スキルの内容に疑問を持ったとき、スキルの改良を提案するときに使う。

どのスキルを導入するか決める。全員が入れる既定のスキル、作業に応じて追加するスキル、npx skills によるホームへの導入と削除、統合済みの旧スキルを入れない判断、リポジトリ固有のスキルを置くかどうかの判断を定義する。スキルを初めて導入するとき、philtzjp/skills に新しいスキルが追加されたとき、作業に必要なスキルが入っていないとき、不要になったスキルを外すときに使う。

Detect software architecture bad smells, algorithmic complexity hotspots, and anti-patterns in a codebase. Produces a detailed markdown report identifying violations of architectural principles, design patterns, code quality, and performanc

Draw how any system moves between states, such as a product flow, a protocol, or a service, as a Mermaid statechart from a TLA+ model. Each bug is a numbered path with blue borders to a state that breaks a rule. Use for /statechart, to show

Render the state of the work in flight as an ASCII status board in the terminal, instead of a status paragraph.

Use when the user wants to mint, view, or manage StockerPunk NFTs on BNB Chain, or check their mining performance. StockerPunk is an agent-only pixel-punk collection where each NFT mines real tokenized stock (NVDA) into its own ERC-6551 wal

Strip explanatory copy out of user interfaces. Use when writing, editing, or reviewing any UI text, including headings, empty states, button labels, toggle and setting descriptions, helper text, dialogs, banners, permission notices, and onb

Use when integrating with Tesla's official Fleet API to read vehicle/energy device data or issue remote commands (e.g. start HVAC preconditioning, wake vehicle, charge controls). Covers onboarding (developer app registration, regions/base U

Create and manage test data with factory patterns, fixture strategies, data anonymization, and synthetic data generation. Covers Fishery (TypeScript), FactoryBot (Ruby), Factory Boy (Python), database seeding, cleanup strategies, and GDPR-c

Design environment strategy for testing across dev, CI, preview, staging, and production — Docker Compose test infrastructure, multi-stage Dockerfiles, seed-data lifecycle, per-PR preview environments, production parity, and external-depend

Shrinking glTF assets for three.js with gltf-transform: inspect, validate and optimize commands, Draco, meshopt and quantization, KTX2, WebP and AVIF textures, the loader setup each output needs, and gltfjsx for React. Use when compressing,

Measuring and optimizing three.js speed: render statistics, memory counters, fewer draw calls via merging, instancing, batching and LOD, fill rate, shadows, CPU work per frame, shader stalls and GPU compute for particles, WebGL or WebGPU wi

Building three.js scenes from React with React Three Fiber and drei: renderer setup, the scene graph as JSX, frame-loop, state and loading hooks, pointer events, and adaptive performance, on WebGL or with WebGPU and TSL. Use when writing Re

Generate a design document (design proposal) from a PRD, in the style of Go's official design proposals — Abstract / Background / Design / Rationale / Compatibility / Implementation, heavy on the 'why' and tradeoffs. Triggers on: to-design,

Decompose a PRD and/or SPEC into implementable, vertically-sliced Issues with real blocking edges, then create them in your chosen platform (GitHub or Local). Use after /prd (and optionally /prd-to-spec) to turn requirements into agent-read

Use to report how hardcoded style values on a page, file or diff map onto a project's existing design tokens, such as "are we using our tokens here", "which hardcoded colors map to which tokens", a token audit or planning a token migration.

Orchestrates modular PyTorch profiler trace analysis with TraceLens: generates perf reports, prepares category data, runs system-level and compute-kernel subagents in parallel, validates outputs, and writes a prioritized stakeholder report

Make any UI appear as if an AI agent is generating it. A dotted wireframe is drawn from the element's own structure (outline, boxes, text lines, icon paths, with guides, corner handles and size labels), then the real UI is revealed behind a

Turborepo と pnpm でモノレポを構成・運用する。apps と packages の分離、workspace:* による内部依存、共有 tsconfig、責務パッケージ、turbo.json のタスク定義とキャッシュ、pnpm コマンドによる package.json 生成、Cargo や Go など非 JavaScript プロジェクトの取り込み。パッケージの追加、turbo.json や pnpm-workspace.yaml や tsconfig の編集、a

パッケージの追加、turbo.json、pnpm-workspace.yaml、tsconfig の編集、apps や packages の構成変更を行うときに参照する。このスキルは turborepo スキルに統合した。内容は turborepo スキルを使う。

Use to critique a screen, flow, prototype or running build, accessibility included, and rank what is broken, such as "check this page before I ship", "is this flow ready", "is it accessible", keyboard or screen-reader checks, "it feels off

Review and understand freshly AI-generated code. Scans the current repo's uncommitted/branch changes and builds a Claude-style light-themed two-column webpage — left is the project folder tree of changed files, right is the selected file's

Generate a Phase-2 Walkthrough artifact (walkthrough.md) once implementation and verification are complete. Captures a Change Summary, Verification Steps (commands + unit tests + automated browser testing outcomes), Visual Proof (screenshot

企业微信 PPT 技能。三类能力：(1) 生成 PPT——根据主题、大纲、文档或参考材料创建在线演示文稿；(2) 编辑在线 PPT；(3) 导入 PPT——把本地 .ppt/.pptx 导入为企微在线文档。生成与编辑一律作用于企微在线 slide 文档，交付形态为在线文档链接。当用户提到'做个PPT'、'生成演示文稿'、'写份汇报PPT'、'把这份文档转成PPT'、'照着这个模板做几页'、'改第3页标题'、'把这页图表数据换掉'、'加一页总结'、'封面换个配色'、'导入PPT

文章を書く・推敲するときに読む。ドキュメント、コメント、コミットメッセージ、PR 本文、記事のような説明文と、小説のような読み物が対象である。「AI っぽい」「読みにくい」文章を直すときにも使う。

Write Celigo settings forms -- the form definition (settingsForm) the Celigo UI renders to collect a resource's settings, on integrations, flow groups, flows, exports, imports, connections, and iClients. Covers the seven scopes and how step