unsloth-kernel-optimization
Optimize and implement GPU kernels using CubeCL for memory-efficient LLM training
Discover DevOps skills, compare source repositories and read the instructions for each workflow.
Optimize and implement GPU kernels using CubeCL for memory-efficient LLM training
Automates GitHub workflows including batch issue creation from JSON/CSV, PR creation with issue linking, pre-PR testing, and WordPress-specific validation (PHP linting, WPCS). Use when creating multiple issues,...
Ensures ASCII/box-drawing diagrams follow consistent formatting standards.
How to organize, structure, and keep a project plan up to date.
Steps to complete prior to creating a pull request.
Provides project status by reading the PLAN.md file. Use this when asked about project progress or status.
Extract content from documents (PDF, Word, Excel, PowerPoint) and answer questions with source citations
Execute Python code in an isolated environment for calculations, file generation, and data processing
Conduct comprehensive web research with source fetching, content extraction, and synthesis into structured reports
Generates session summaries and resume prompts for multi-session work. Use when completing features, before context limits (~50% capacity), or when user says "summary", "wrap up", "save progress", "end session"...
Set up Ralph for autonomous feature development. Use when starting a new feature that Ralph will implement. Triggers on: ralph, set up ralph, start ralph, new ralph feature, ralph setup. Chats through the featu...
Generate a Product Requirements Document (PRD) for a new feature. Use when planning a feature, starting a new project, or when asked to create a PRD. Triggers on: create a prd, write prd for, plan this feature,...
Set up Ralph for autonomous feature development. Use when starting a new feature that Ralph will implement. Triggers on: ralph, set up ralph, start ralph, new ralph feature, ralph setup. Chats through the featu...
Browser automation with persistent page state. Use when users ask to navigate websites, fill forms, take screenshots, extract web data, test web apps, or automate browser workflows. Trigger phrases include "go...
Automatically maintain documentation-as-memory in AI-assisted development projects. This skill detects repos using the ai-doc-first-framework structure and maintains PROGRESS.md, DECISIONS.md, ARCHITECTURE.md,...
OpenAI Codex CLIを使用したコードレビュー、分析、コードベースへの質問を実行する。使用場面: (1) コードレビュー依頼時、(2) コードベース全体の分析、(3) 実装に関する質問、(4) バグの調査、(5) リファクタリング提案、(6...
Codex CLI(read-only)を用いて、レビュー→Claude Code修正→再レビュー(ok: true まで)を反復し収束させるレビューゲート。仕様書/SPEC/PRD/要件定義/設計、実装計画(PLANS.md等)の作成・更新直後、major step(>=5 files...
Converts PRD markdown files to Ralph-compatible prd.json format for autonomous execution. Use when asked to convert a PRD, create prd.json, or prepare stories for Ralph.
Creates Product Requirements Documents (PRDs) for features with user stories in Ralph-compatible format. Use when asked to create a PRD, write requirements, plan a feature, or generate user stories.
Comprehensive knowledge base for Photometrics AI, a street lighting optimization startup. Use when Ari needs help with any Photometrics AI business activity including sales materials, proposals, presentations,...
Render Manim scenes to still frames and review them to diagnose animation/layout issues. Use when debugging scenes, verifying visual correctness, or inspecting frame-by-frame output.
Guide for customizing padz CLI output using templates and stylesheets. Use when working on padz rendering, modifying how pads are displayed, changing colors/styles, editing Jinja templates, or working with the...
Guide for customizing padz CLI output: templates, styles, and rendering. Use when working on: (1) Modifying templates in crates/padz/src/cli/templates/, (2) Adding or changing styles in styles/default.yaml, (3)...
Explains the padz hybrid store architecture—files are truth, JSON metadata is a rebuildable cache. Covers self-healing reconciliation (orphans, zombies, staleness), the Backend+Store split for testability, and...
The default order uses GitHub repository stars from our indexed data. Stars belong to the source repository and may be shared by multiple skills. They indicate repository interest, not skill installs or an independent quality score.