Overview CLI FAQ

Agent Skills 排行榜 · 关键词 + 语义搜索

按 / 聚焦搜索框
正在使用 AI 进行语义搜索...
24,815
总 Skills
98.6M
总安装量
2,594
贡献者
# Skill 仓库 描述 安装量
9101 comic-panel-generation eachlabs/skills
Comic Panel Generation Generate professional comic and manga artwork using each::sense. This skill creates single panels, multi-panel strips, full page layouts, and complete comic sequences with consistent character designs. Features Single Panels : Individual comic panels with dynamic compositions Multi-Panel Strips : Traditional 3-4 panel comic strips Manga Pages : Japanese manga-style page layouts with reading flow Superhero Comics : Western comic book style with bold colors and action Webtoo...
 586
9102 android-adb httprunner/skills
Android ADB Reference for controlling Android devices with raw adb commands. Execution Constraints Use adb -s <serial> whenever more than one device is connected. Confirm screen resolution before coordinate actions: adb -s SERIAL shell wm size . Ask for missing required inputs before executing (serial, package/activity, coordinates, APK path). Surface actionable stderr on failure (authorization, cable/network, tcpip state). Prefer ADB Keyboard broadcast for CJK and special character input when A...
 586
9103 muapi-instagram-post samuraigpt/generative-media-skills
Instagram Post Create a polished, on-brand Instagram post — square or portrait hero image with matching caption and hashtags. Inputs Name Type Required Default Description brief text yes — What the post is about (e.g. "summer coffee launch at our café, warm golden vibes"). brand_style text no modern, vibrant, clean typography, lifestyle photography aesthetic Brand personality and visual style tags. format text no 1:1 Post format — "1:1" for feed square, "4:5" for portrait feed, "9:16" for Reels....
 586
9104 resolve-conflicts antinomyhq/forge
Git Conflict Resolution Resolve Git merge conflicts by intelligently combining changes from both branches while preserving the intent of both changes. This skill follows a plan-first approach: assess conflicts, create a detailed resolution plan, get approval, then execute. Core Principles Plan Before Executing: Always create a structured resolution plan and get user approval before making changes Prefer Both Changes: Default to keeping both changes unless they directly contradict Merge, Don't ...
 586
9105 compose-state-holder-ui-split chrisbanes/skills
Compose: state holder/UI split Core principle Separate state-holder wiring from UI rendering. The state-holder composable talks to ViewModels, components, flows, navigation, and side effects. The UI composable takes plain immutable UI state plus callbacks and describes layout. This keeps screens previewable, testable, and easier to reuse across Android, Desktop, TV, and KMP/CMP targets. When to use this skill Use this when a Compose screen: Takes a ViewModel, component, controller, navigator, re...
 586
9106 nemo-mbridge-perf-memory-tuning nvidia/skills
Memory Tuning Stable docs: @docs/parallelisms.md Card: @skills/nemo-mbridge-perf-memory-tuning/card.yaml What It Is GPU OOM failures during training often stem from memory fragmentation rather than raw capacity. PyTorch's default CUDA allocator can leave unusable gaps between allocations. The single most effective fix is: export PYTORCH_CUDA_ALLOC_CONF = expandable_segments:True This tells PyTorch to use expandable (non-fixed-size) memory segments, which dramatically reduces fragmentation and ...
 586
9107 gws-cloudidentity googleworkspace/cli
cloudidentity (v1) PREREQUISITE: Read ../gws-shared/SKILL.md for auth, global flags, and security rules. If missing, run gws generate-skills to create it. gws cloudidentity < resource > < method > [ flags ] API Resources customers userinvitations — Operations on the 'userinvitations' resource devices cancelWipe — Cancels an unfinished device wipe. This operation can be used to cancel device wipe in the gap between the wipe operation returning success and the device being wiped. This operation is...
 585
9108 cursor-best-practices siviter-xyz/dot-agent
Cursor Best Practices Best practices for working with Cursor to maximize productivity and code quality. Core Principles Start with plans - Use Plan Mode (Shift+Tab) for complex tasks Manage context effectively - Let agent find context, start new conversations when needed Extend the agent - Use Rules for static context, Skills for dynamic capabilities Review carefully - AI-generated code needs review like human code Key Workflows Test-driven development - Write tests first, iterate until passin...
 585
9109 nextjs-best-practices davila7/claude-code-templates
Next.js Best Practices Principles for Next.js App Router development. 1. Server vs Client Components Decision Tree Does it need...? │ ├── useState, useEffect, event handlers │ └── Client Component ('use client') │ ├── Direct data fetching, no interactivity │ └── Server Component (default) │ └── Both? └── Split: Server parent + Client child By Default Type Use Server Data fetching, layout, static content Client Forms, buttons, interactive UI 2. Data Fetching Patterns Fetch Strategy Pattern Us...
 585
9110 autonomous camacho/ai-skills
Return to Autonomous Mode Claude Code only — Codex and Cursor sessions are always autonomous. Activate .claude/hooks/activate-autonomous.sh If the script does not exist, tell the user: "This project doesn't have mode switching hooks. You are already in autonomous mode by default." Confirmation After activation, print: Mode: autonomous Worktree enforcement active. Full workflow pipeline. Behavior Follow autonomous mode rules in .claude/rules/operating-mode.md .
 585
9111 react-refactor pproenca/dot-skills
React Refactor Best Practices Architectural refactoring guide for React applications. Contains 40 rules across 7 categories, prioritized by impact from critical (component and state architecture) to incremental (refactoring safety). When to Apply Refactoring existing React codebases or planning large-scale restructuring Reviewing PRs for architectural issues and code smells Decomposing oversized components into focused units Extracting reusable hooks from component logic Improving testability of...
 585
9112 deepinit yeachan-heo/oh-my-claudecode
Deep Init Skill Creates comprehensive, hierarchical AGENTS.md documentation across the entire codebase. Core Concept AGENTS.md files serve as AI-readable documentation that helps agents understand: What each directory contains How components relate to each other Special instructions for working in that area Dependencies and relationships Hierarchical Tagging System Every AGENTS.md (except root) includes a parent reference tag: <!-- Parent: ../AGENTS.md --> This creates a navigable hierar...
 585
9113 hyva-ui-component hyva-themes/hyva-ai-tools
Hyva UI Component Applies Hyva UI template-based (non-CMS) components to a Hyvä theme by copying files from {hyva_ui_path}/components/ to app/design/frontend/{Vendor}/{Theme}/. Path variable: {hyva_ui_path} = vendor/hyva-themes/hyva-ui (default) or user-provided custom path. Command execution: For any commands that need to run inside the development environment (e.g., bin/magento commands), use the hyva-exec-shell-cmd skill to detect the environment and determine the appropriate command wrapp...
 585
9114 subtitle-generation eachlabs/skills
Subtitle Generation Generate professional subtitles and captions for videos using each::sense. This skill creates accurate transcriptions, multi-language subtitles, animated captions, and various export formats optimized for social media, video production, and accessibility. Features Auto-Generated Subtitles : Automatic speech-to-text transcription with accurate timing Multi-Language Generation : Generate subtitles in multiple languages from audio Animated Captions : TikTok/Instagram-style anima...
 585
9115 langfuse sickn33/antigravity-awesome-skills
Langfuse This skill helps you use Langfuse effectively across all common workflows: instrumenting applications, migrating prompts, debugging traces, and accessing data programmatically. Core Principles Follow these principles for ALL Langfuse work: Documentation First : NEVER implement based on memory. Always fetch current docs before writing code (Langfuse updates frequently) See the section below on how to access documentation. CLI for Data Access : Use langfuse-cli when querying/modifying Lan...
 585
9116 shadcn vercel/vercel-plugin
shadcn/ui A framework for building ui, components and design systems. Components are added as source code to the user's project via the CLI. IMPORTANT: Run all CLI commands using the project's package runner: npx shadcn@latest , pnpm dlx shadcn@latest , or bunx --bun shadcn@latest — based on the project's packageManager . Examples below use npx shadcn@latest but substitute the correct runner for the project. Current Project Context !`npx shadcn@latest info --json 2 >/dev/ null || echo ' { "error...
 585
9117 kanchi-dividend-us-tax-accounting tradermonty/claude-trading-skills
Kanchi Dividend Us Tax Accounting Overview Apply a practical US-tax workflow for dividend investors while keeping decisions auditable. Focus on account placement and classification, not legal/tax advice replacement. When to Use Use this skill when the user needs: US dividend tax classification planning (qualified vs ordinary assumptions). Holding-period checks before year-end tax planning. Account-location decisions for stock/REIT/BDC/MLP income holdings. A standardized annual dividend tax memo ...
 585
9118 dynamo-troubleshoot nvidia/skills
Dynamo Troubleshoot Purpose Turn a Dynamo failure into a clear problem class, strongest signal, and next action. Start with read-only evidence, avoid secrets, and fix one layer at a time. Prerequisites Python 3.10+ on the operator machine. kubectl configured with read access to the target namespace. Permission to read pods, events, jobs, PVCs, and DynamoGraphDeployment resources (NOT secrets). Network reachability to the cluster API server. Show more Installs 549 Repository nvidia/skills GitHub ...
 585
9119 project-planning jezweb/claude-skills
Project Planning Skill Specialized planning assistant for web application projects. Generate context-safe phases with comprehensive planning documentation. ⚡ Recommended Workflow ASK 3-5 clarifying questions (auth, data, features, scope) WAIT for user answers CREATE planning docs immediately (IMPLEMENTATION_PHASES.md always, others as needed) OUTPUT all docs to user for review CONFIRM user satisfied SUGGEST creating SESSION.md and starting Phase 1 🤖 Automation Commands Two slash commands are ...
 584
9120 stock-quote octagonai/skills
Stock Quote Retrieve real-time stock quotes with current price, volume, day range, 52-week range, market cap, and moving averages using the Octagon MCP server. Prerequisites Ensure Octagon MCP is configured in your AI agent (Cursor, Claude Desktop, Windsurf, etc.). See references/mcp-setup.md for installation instructions. Workflow 1. Identify the Stock Determine the ticker symbol for the stock you want to quote (e.g., AAPL, MSFT, GOOGL). 2. Execute Query via Octagon MCP Use the octagon-agent to...
 584
9121 brewpage-publish kochetkov-ma/claude-brewcode
brewpage Publish content to brewpage.app — free instant hosting for HTML pages, JSON documents, and files. No sign-up required. Workflow Step 1: Parse Arguments Extract from $ARGUMENTS : --ttl N → TTL in days (default: 5 ) Remaining text → content_arg Step 2: Detect Content Type Input Type API content_arg is a path AND file exists ( test -f ) FILE POST /api/files (multipart) content_arg starts with { or [ JSON POST /api/json Anything else HTML POST /api/html (format=markdown) For FILE: get file ...
 584
9122 mongoose-mongodb pluginagentmarketplace/custom-plugin-nodejs
Mongoose & MongoDB Skill Master MongoDB database integration in Node.js with Mongoose, the elegant object modeling library. Quick Start Connect and CRUD in 4 steps: Install - npm install mongoose Connect - mongoose.connect(uri) Define Schema - Create data models CRUD - Create, Read, Update, Delete Core Concepts Connection Setup const mongoose = require('mongoose'); mongoose.connect(process.env.MONGODB_URI, { useNewUrlParser: true, useUnifiedTopology: true }); mongoose.connection.on('co...
 584
9123 security-patterns yonatangross/orchestkit
Security Patterns Comprehensive security patterns for building hardened applications. Each category has individual rule files in rules/ loaded on-demand. Quick Reference Category Rules Impact When to Use Authentication 3 CRITICAL JWT tokens, OAuth 2.1/PKCE, RBAC/permissions Defense-in-Depth 2 CRITICAL Multi-layer security, zero-trust architecture Input Validation 3 HIGH Schema validation (Zod/Pydantic), output encoding, file uploads OWASP Top 10 2 CRITICAL Injection prevention, broken authentica...
 584
9124 hyva-theme-list hyva-themes/hyva-ai-tools
Hyvä Theme Listing Lists all Hyvä theme paths in a Magento 2 project. Themes are identified by the presence of web/tailwind/package.json. Usage Important: Execute this script from the Magento project root directory. Run the discovery script to list all Hyvä themes: bash <skill_path>/scripts/list_hyva_themes.sh Where <skill_path> is the directory containing this SKILL.md file (e.g., .claude/skills/hyva-theme-list). Output format: One theme path per line (relative to project root), or empt...
 584
9125 hyva-render-media-image hyva-themes/hyva-ai-tools
Hyvä Render Image Generate responsive <picture> elements for Hyvä Theme templates using the \Hyva\Theme\ViewModel\Media view model. When to Use Adding images to Hyvä PHTML templates Creating responsive images with different sources for mobile/desktop Implementing hero banners, product images, or CMS content images Optimizing images for Core Web Vitals (LCP, CLS) Workflow 1. Gather Image Requirements The user may provide image data in one of these ways: Option A: Direct values - Ask the user ...
 584
9126 deployment-engineer charon-fan/agent-playbook
Deployment Engineer Specialist in deployment automation, CI/CD pipelines, and infrastructure management. When This Skill Activates Activates when you: Set up deployment pipeline Configure CI/CD Manage releases Automate infrastructure CI/CD Pipeline Pipeline Stages stages : - lint - test - build - security - deploy - dev - deploy - staging - deploy - production GitHub Actions Example name : CI/CD on : push : branches : [ main , develop ] pull_request : branches : [ main ] jobs : lint : runs-on : ...
 584
9127 grepai-search-boosting yoanbernabeu/grepai-skills
This skill covers configuring score boosting to prioritize relevant code paths and deprioritize tests, docs, and vendor code. When to Use This Skill - Prioritizing source code over tests - Penalizing vendor/third-party code - Boosting important directories - Customizing result ranking What is Boosting? Boosting modifies search scores based on file paths: ``` Original score: 0.85 (src/auth.go) Bonus (+10%): 0.935 Original score: 0.85 (tests/auth_test.go) Penalty (-50%): 0.425 ``` T...
 584
9128 youtube-transcript michalparkola/tapestry-skills-for-claude-code
YouTube Transcript Extract transcripts from YouTube videos using the youtube-transcript-api. Usage Run the script with a YouTube URL or video ID: uv run scripts/get_transcript.py "VIDEO_URL_OR_ID" With timestamps: uv run scripts/get_transcript.py "VIDEO_URL_OR_ID" --timestamps Defaults Without timestamps (default): Plain text, one line per caption segment With timestamps: [MM:SS] text format (or [HH:MM:SS] for longer videos) Supported URL Formats https://www.youtube.com/watch?v=VIDEO_ID ...
 584
9129 edge-strategy-designer tradermonty/claude-trading-skills
Edge Strategy Designer Overview Translate concept-level hypotheses into concrete strategy draft specs. This skill sits after concept synthesis and before pipeline export validation. When to Use You have edge_concepts.yaml and need strategy candidates. You want multiple variants (core/conservative/research-probe) per concept. You want optional exportable ticket files for interface v1 families. Prerequisites Python 3.9+ PyYAML edge_concepts.yaml produced by concept synthesis Output strategy_drafts...
 584
9130 cupynumeric-parallel-data-load nvidia/skills
Parallel sharded data -> cupynumeric load Why this skill exists. cupynumeric mirrors NumPy's array API, including cupynumeric.load for a single .npy file. Beyond that, file loading lives in Legate, not cupynumeric: Format Built-in loader Single .npy cupynumeric.load(path) (NumPy-API parity) HDF5 (single file) legate.io.hdf5.from_file / from_file_batched Sharded multi-file (any format), Parquet/Arrow, raw binary, custom layouts No built-in loader — this skill. This skill shows the canonical way t...
 584
9131 pi-cli-runtime agents365-ai/365-skills
Pi Runtime Use this skill only inside the pi:pi-rescue subagent. Primary helper: node "${CLAUDE_PLUGIN_ROOT}/scripts/pi-companion.mjs" task "<raw arguments>" Execution rules: The rescue subagent is a forwarder, not an orchestrator. Its only job is to invoke task once and return that stdout unchanged. Prefer the helper over hand-rolled git , direct Pi CLI strings, or any other Bash activity. Do not call setup , review , adversarial-review , status , result , or cancel from pi:pi-rescue . Use task...
 584
9132 copilotkit-integrations copilotkit/copilotkit
CopilotKit Integrations Live Documentation (MCP) This plugin includes an MCP server ( copilotkit-docs ) that provides search-docs and search-code tools for querying live CopilotKit documentation and source code. Useful for looking up framework-specific integration details. Claude Code: Auto-configured by the plugin's .mcp.json -- no setup needed. Codex: Requires manual configuration. See the copilotkit-debug skill for setup instructions. Overview CopilotKit connects to external agent frameworks ...
 584
9133 longbridge-kline longbridge/skills
longbridge-kline Historical candlesticks and today's intraday curve for Longbridge-supported securities (HK / US / A-share / Singapore). Does not support options, warrants, or indices — defer to longbridge-derivatives (derivatives) or longbridge-quote (indices). Response language : match the user's input language — Simplified Chinese / Traditional Chinese / English. Subcommands Run longbridge kline --help to confirm current period values, defaults, and aliases. CLI command Use when longbridge kl...
 583
9134 secrets-vault-manager alirezarezvani/claude-skills
Secrets Vault Manager Tier: POWERFUL Category: Engineering Domain: Security / Infrastructure / DevOps Overview Production secret infrastructure management for teams running HashiCorp Vault, cloud-native secret stores, or hybrid architectures. This skill covers policy authoring, auth method configuration, automated rotation, dynamic secrets, audit logging, and incident response. Distinct from env-secrets-manager which handles local .env file hygiene and leak detection. This skill operates at the ...
 583
9135 feishu-cli-bitable riba2534/feishu-cli
飞书多维表格（Bitable / Base） 通过 base/v3 API 操作飞书多维表格。 bitable 也支持 base 别名。 API 切换 ：此技能已从旧的 bitable/v1 切换到新的 base/v3 。字段名 app_token 和 base_token 在飞书文档里是同一个值的两种叫法（老 v1 叫 app_token，新 v3 叫 base_token），CLI 只认 --base-token （ --app-token 已删除）。从多维表格 URL 里 /base/<token> 或 /bitable/<token> 片段里取 token 即可。 前置条件 认证 ：所有命令默认使用 User Access Token （执行 feishu-cli auth login 登录） App 凭证 ：应用 App ID + App Secret（base/v3 需要 X-App-Id header，自动注入） 命令速查 基础（3 命令） 创建多维表格 feishu-cli bitable create --name "项目管理" --time-zone Asia/S...
 583
9136 python-testing mindrally/skills
Python Testing Patterns Comprehensive testing strategies for Python applications using pytest, TDD methodology, and best practices. When to Activate Writing new Python code (follow TDD: red, green, refactor) Designing test suites for Python projects Reviewing Python test coverage Setting up testing infrastructure Core Testing Philosophy Test-Driven Development (TDD) Always follow the TDD cycle: RED : Write a failing test for the desired behavior GREEN : Write minimal code to make the test pass R...
 583
9137 game-audio opusgamelabs/game-creator
Game Audio Engineer (Web Audio API) You are an expert game audio engineer. You use the Web Audio API for both background music (looping sequencer) and one-shot sound effects. Zero dependencies — everything is built into the browser. Performance Notes Take your time with each step. Quality is more important than speed. Do not skip validation steps — they catch issues early. Read the full context of each file before making changes. Test every sound in the browser. Web Audio timing is different fro...
 583
9138 aws-skills sickn33/antigravity-awesome-skills
Aws Skills Overview AWS development with infrastructure automation and cloud architecture patterns When to Use This Skill Use this skill when you need to work with aws development with infrastructure automation and cloud architecture patterns. Instructions This skill provides guidance and patterns for aws development with infrastructure automation and cloud architecture patterns. For more information, see the source repository .
 583
9139 anki-connect intellectronica/agent-skills
AnkiConnect Overview Enable reliable interaction with Anki through the AnkiConnect local HTTP API. Use this skill to translate user requests into AnkiConnect actions, craft JSON requests, run them via curl/jq (or equivalent tools), and interpret results safely. Preconditions and Environment If Anki is not running, launch Anki, then wait until the AnkiConnect server responds at http://127.0.0.1:8765 (default). Verify readiness using curl, e.g. curl -sS http://127.0.0.1:8765 should return Anki-C...
 583
9140 skill-security-auditor alirezarezvani/claude-skills
Skill Security Auditor Scan and audit AI agent skills for security risks before installation. Produces a clear PASS / WARN / FAIL verdict with findings and remediation guidance. Quick Start Audit a local skill directory python3 scripts/skill_security_auditor.py /path/to/skill-name/ Audit a skill from a git repo python3 scripts/skill_security_auditor.py https://github.com/user/repo --skill skill-name Audit with strict mode (any WARN becomes FAIL) python3 scripts/skill_security_auditor.py /path...
 583
9141 chinese-commit-conventions jnmetacode/superpowers-zh
中文 Git 提交规范 1. Conventional Commits 中文适配 基于 Conventional Commits 1.0.0 规范，针对中文团队的实际使用习惯进行适配。 类型（type）定义 Show more Installs 423 Repository jnmetacode/supe…owers-zh GitHub Stars 4.5K First Seen Mar 21, 2026 Security Audits Gen Agent Trust Hub Pass Socket Pass Snyk Pass
 583
9142 close-month anthropics/knowledge-work-plugins
Run the month-end close workflow. Reconcile, flag gaps, narrate the P&L, and export the close packet for the owner's records (and their accountant). Parse arguments: --month (default: previous calendar month) — YYYY-MM format --save-to (default files ) — files (Google Drive / OneDrive), desktop (local), or both Step 1 — Reconcile Trigger the month-end-prep skill workflow: Pull all QuickBooks transactions for the target month. Pull settlements from each connected payment processor (PayPal, Stripe...
 583
9143 market-regimes 0xhubed/agent-trading-arena
Market Regimes Last updated: 2026-01-17 20:31 UTC Active patterns: 32 Total samples: 0 Confidence threshold: 60% How to Use This Skill Identify the current market regime using price action and volatility Look up the recommended strategy for that regime below Adjust your trading approach accordingly Monitor for regime changes Regime Strategies Mixed Choppy Recommended approach (93% confidence, seen 1x): Reduce or eliminate trading. In mixed markets (BNB +0.93%, SOL +1.65%, BTC -0.08%, ETH -0....
 582
9144 gcp-cloud-architect alirezarezvani/claude-skills
GCP Cloud Architect Design scalable, cost-effective Google Cloud architectures for startups and enterprises with infrastructure-as-code templates. Workflow Step 1: Gather Requirements Collect application specifications: Show more
 582
9145 electron-development mindrally/skills
Electron Development Guidelines You are an expert in Electron development for building cross-platform desktop applications. Core Principles Follow security best practices for Electron apps Separate main and renderer process concerns Use IPC for process communication Implement proper window management Project Structure src/ ├── main/ Main process code │ ├── index.ts Entry point │ ├── ipc/ IPC handlers │ └── utils/ Utilities ├── renderer/ Ren...
 582
9146 landing-page-generator alirezarezvani/claude-skills
Pages: Landing Page Guides campaign landing page structure, conversion flow, and optimization. Primary use : Paid ads (PPC/SEM) — landing pages are typically built to receive paid traffic; ad-to-page alignment is critical for conversion. See paid-ads-strategy for when to use paid ads and ad-to-page alignment principles. Also applies to affiliate signup, product launch, lead capture, webinar registration, and other single-goal conversion pages. Scale : When building many landing pages (city-speci...
 582
9147 launch-nemo-rl nvidia/skills
launch-nemo-rl — running NeMo-RL recipes on Kubernetes via nrl-k8s This is the playbook for the nrl-k8s CLI at infra/nrl_k8s/ . Follow it when the user asks to launch / iterate / debug a NeMo-RL recipe on a Kubernetes cluster. Verify current state ( kubectl , git log , the recipe + infra files) before acting — the cluster is shared and the cost of a wrong action is high. 1. One command, two modes There is a single top-level submission command: nrl-k8s run . It has two lifecycle modes. Mode Invoc...
 582
9148 nemo-mbridge-perf-cuda-graphs nvidia/skills
CUDA Graphs Stable documentation: @docs/training/cuda-graphs.md Card: @skills/nemo-mbridge-perf-cuda-graphs/card.yaml What It Is CUDA graphs capture GPU operations once and replay them with minimal host-driver overhead. Bridge supports two implementations: cuda_graph_impl Mechanism Scope support "local" MCore FullCudaGraphWrapper wrapping entire fwd+bwd full_iteration "transformer_engine" TE make_graphed_callables() per layer attn , mlp , moe , moe_router , moe_preprocess , mamba Quick Decision ...
 582
9149 nemo-mbridge-multi-node-slurm nvidia/skills
Multi-Node Slurm Convert single-node uv run python -m torch.distributed.run commands into multi-node Slurm sbatch scripts with Enroot container support, and debug common multi-node failures. Two Approaches: srun-native vs uv run torch.distributed Approach ntasks-per-node Process spawning Best for srun-native (preferred) 8 Slurm spawns 8 tasks/node Conversion, inference, Bridge scripts uv run torch.distributed (legacy) 1 uv run python -m torch.distributed.run spawns 8 procs/node MLM pretrain_gpt....
 582
9150 cupynumeric-hdf5 nvidia/skills
cuPyNumeric HDF5 I/O Purpose Use legate.io.hdf5 to read and write cuPyNumeric arrays as HDF5 files. Reach for it whenever a cuPyNumeric array must land in — or load from — an .h5 / .hdf5 file: every rank reads and writes its own tile in parallel, so never funnel a large array through a single process. Answer inline. Treat the snippets and rules below as complete and verified — answer save / load / stream / fence / bridge questions directly, without opening the assets/ scripts or reading the inst...
 582