alexanderop/aop-mode

記録された最新のソース活動: 2026年10月5日 05:45

SkillsMP カタログ更新: 2026年11月12日 12:52

収集済み skills: 12

GitHub スター: 0
GitHub フォーク: 0

AI アシスタントでインストール

このプロンプトをコピーして、使用中の AI アシスタントに送ってください。

Follow https://skillsmp.com/skill-install/prompt.md to install Agent Skills from https://github.com/alexanderop/aop-mode.

リポジトリの README

README の元ファイル：README.md

README の元ファイルを読む

README 全文を表示

# aop-mode [Documentation](https://alexanderop.github.io/aop-mode/) The complete original **[pstack by Lauren Tan](https://github.com/cursor/plugins/tree/main/pstack)** skill set, packaged for explicit use in Claude Code, Codex, and GitHub Copilot CLI. Independent project, not a GitHub fork or official port. See [NOTICE.md](NOTICE.md). All **50 main skills + 3 automation skills**, **23 playbooks**, two agent prompts, and every supporting file are pinned to pstack v0.15.9 at `e43c7ee26e0038c6c1fa8380dd34ce86ff94cb2a`. The 161 original files remain unchanged. `aop-mode` adds an entrypoint to the full poteto-mode router. No startup hooks. ## Install the plugin Ready-built packages contain all skills and require no pnpm on the user's machine. The verification workflow produces the `aop-mode-plugins` download artifact. See [plugin installation](https://alexanderop.github.io/aop-mode/plugin-installation/) for Claude Code, Codex, and Copilot CLI instructions. To build an archive locally: ```sh pnpm install --frozen-lockfile pnpm plugins:pack ``` Extract `dist/releases/aop-mode-0.3.0.tar.gz` and follow its `INSTALL.md`. Every client package has a standard Agent Plugins manifest; the Claude package also carries its compatibility manifest. The archive includes local marketplace catalogs, attribution, and a SHA-256 checksum alongside it. ```sh pnpm eval:plugins # isolated CLI install, discovery, update and removal; no model calls ``` ## Install project-local skills ```sh pnpm install --frozen-lockfile pnpm skills:install codex /absolute/path/to/your-project pnpm dev ``` Use `claude` or `copilot` for the other targets. Installation copies all 63 entrypoints and the complete upstream tree into project-local skill directories. Existing skill-name collisions abort before writing anything. No global settings are changed. Invoke `$aop-mode` in Codex or `/aop-mode` in Claude Code/Copilot CLI. Individual skills retain their original names: `$tdd`, `/interrogate`, `/how`, etc. For headless Copilot, explicitly request reading the installed SKILL.md. Use `$asd` to restate the previous answer in simplified technical language. Use `$visual` (or `/visual` in Claude Code/Copilot CLI) to turn an answer or topic into a dark HTML explainer and open it in the browser. For example: `$visual explain how this project's evaluation runner works`. The visual skill uses `asd` for writing and includes an article template with diagrams, a contents tree, and Shiki code highlighting through a pinned CDN module. The article remains readable offline; syntax highlighting requires network access. Use `$improve-codebase-architecture` to survey an existing codebase and open a visual report of ranked refactoring opportunities. Add an area to narrow the review, for example `$improve-codebase-architecture src/runner.ts`. It traces real callers, explains proposed boundaries, and identifies migration risks and verification needs. Application code changes require an implementation request. ## Verify and build ```sh pnpm verify # source integrity, packages, types, lint, tests, docs pnpm test:docs # Playwright documentation journeys pnpm test:harnesses # authenticated CLI task matrix; consumes usage pnpm eval:workflows --list # inspect Codex/Copilot workflow contracts; no model calls pnpm eval:workflows --task investigation # live routing/delegation eval; consumes usage pnpm skills:build # dist/claude, dist/codex, dist/copilot pnpm upstream:check # hashes, file inventory, executable modes pnpm report .eval-artifacts/<run>/summary.json ``` The source import is reproducible with `pnpm exec tsx scripts/import-pstack.ts --source /path/to/cursor-plugins`. `--check --source /path/to/cursor-plugins` also compares the import with Git objects. See [the full catalog](docs/src/content/docs/catalog.md), [compatibility limits](docs/src/content/docs/compatibility.md), and [latest task evidence](docs/src/content/docs/evidence/latest.md). **Content completeness is verified; universal runtime parity is not.** Cursor cloud-agent lifecycle and Automation webhooks are not implemented for other hosts. Multi-model panels, transcript mining, browser controls and external skills depend on available host capabilities. Adapters report missing steps. All source remains included, including make-bot-ui and Benny. Live tests cover repair, read-only review and ordinary-task non-activation, not all 53 workflows or marketplace/IDE/cloud-agent installation. Each run uses a fresh workspace; graders and raw logs remain outside it. Claude requires a working login. Astro Starlight documentation. Strict TypeScript, pnpm, Effect evaluation lifecycle, Vitest and Playwright. Requires Node.js 22.14+, Git and authenticated CLIs for live runs. Original pstack MIT attribution accompanies every distribution. [Workflow evaluations](docs/src/content/docs/workflow-evals.md) inspect actual skill reads, playbook routing, native subagents, verification, and independent outcomes. Compare explicit, installed-inactive, and baseline conditions; missing telemetry remains unobservable. Private per-trial reports link verdicts to native event lines.

リポジトリの README を読み込んでいます…

このリポジトリの skills

分類待ち

未分類

Apply when designing code with time, randomness, persistence, network, or service dependencies, or when tests need to replace them. Make dependencies explicit without unnecessary abstraction.

原文の言語: 英語

2026-10-03

収集済み skill 12 件中 12 件を表示しています。