cpu-mode-assertions-torch
Assert CPU-only runtime inside a container using PyTorch (torch.cuda.is_available()==False) and optional env var checks. Use to prevent accidental GPU execution during CPU smoke tests.
探索开发、内容、数据与办公效率等 AI 技能。先比较用途和仓库热度,再查看技能说明与来源。
Assert CPU-only runtime inside a container using PyTorch (torch.cuda.is_available()==False) and optional env var checks. Use to prevent accidental GPU execution during CPU smoke tests.
Quick scan for version strings across common files (README, pyproject, requirements, package manifests) to avoid release blockers from inconsistent versions. Use before tagging or publishing.
Run analyzer MCP tools (Ruff + optional dead-code scan) as a quality gate before QA/UAT/release.
Run Python pytest suites and capture reproducible evidence (command, environment, output) into agent-output artifacts. Use for QA verification, debugging, and pre-release confidence.
Create comprehensive implementation plans using Codex (GPT 5.2) for complex features. Use when user says "use codex plan mode", "codex plan", "plan with codex", or wants detailed implementation planning with fi...
Patterns for delegating tasks to Codex CLI (GPT 5.2) experts including bug analysis, code review, and plan revision via the mcp__codex__codex tool.
Conduct online research with source attribution, Notion integration, and persistent memory. Organizes findings by project in research/ folder.
TypeScript patterns for types, interfaces, and type guards in viya-app. Use when defining types, working with generated API types, or creating type-safe utilities.
Update and maintain GitHub issues during development. Add progress comments, update project board status, handle blockers, link PRs, and close issues with summaries. Use when user asks to update an issue, chang...
PR workflow, release notes, build management, and GitHub conventions. Use when creating PRs, writing commit messages, or preparing releases.
Start Penpot MCP servers and work with designs. Enables AI-assisted design workflows with Penpot.
Write product strategy documents for Viya TMS product teams. Uses Playing to Win framework, cascading from company strategy to team-level strategic choices.
Headless browser debugging for QA failures and visual testing. Use when investigating UI bugs, taking screenshots, or inspecting DOM and network requests.
Create and improve product documentation for Viya TMS. Supports user guides, feature docs, API references, and release notes with consistent style and structure.
Writing Playwright E2E tests following project patterns and fixtures. Use when creating new E2E tests, debugging test failures, or setting up test helpers and page objects.
Writing clear, user-facing documentation for non-technical users. Use when creating in-app docs, help content, or feature documentation.
Senior engineer code review patterns for Vue/TypeScript projects. Use when reviewing PRs, verifying checklists, or writing release notes.
Understanding viya-app project structure and where to find things. Use when exploring the codebase, finding files, or understanding how the app is organized.
Writing unit tests with vitest and vue-test-utils following project patterns. Use when creating tests for components, composables, utilities, or stores.
MongoDB database development - queries, aggregations, schema analysis using MCP tools
Create and refine Agent Skills following the agentskills.io specification. Use when user asks to write a skill, create a skill, or build automation workflows for AI agents.
Manage viya-app local development environment - set service versions from PRs, restart containers
Extracts and analyzes competitors' ads from ad libraries (Facebook, LinkedIn, etc.) to understand what messaging, problems, and creative approaches are working. Helps inspire and improve your own ad campaigns.
Creating Vue 3 components following project conventions and script order. Use when building new components, refactoring existing ones, or reviewing component code for convention compliance.
默认按已收录数据中的 GitHub 仓库 Stars 排序。Stars 属于来源仓库,同一仓库的多个技能可能共享相同数值;它反映仓库关注度,不代表单个技能的安装量或质量评分。