发现 Skills

Use when ANY Metal porting issue occurs - black screen, rendering artifacts, shader errors, wrong colors, performance regressions, GPU crashes

Use when writing unit tests, adopting Swift Testing framework, making tests run faster without simulator, architecting code for testability, testing async code reliably, or migrating from XCTest - covers @Test/@Suite macros,

Reference — Complete SwiftUI adaptive layout API guide covering ViewThatFits, AnyLayout, Layout protocol, onGeometryChange, GeometryReader, size classes, and iOS 26 window APIs

Manage OmniFocus tasks (macOS only)

Step-by-step guidance for algolia search.

Access ClinPGx pharmacogenomics data (successor to PharmGKB). Query gene-drug interactions, CPIC guidelines, allele functions, for precision medicine and genotype-guided dosing decisions.

Step-by-step guidance for linear issue.

Guidance and answers for mcp scripts.

Archived virtual IR remote executor. Kept as reference; V1 Xiaomi control uses mi-cli.

Web scraping skill using the Firecrawl API for deep content extraction, format conversion, and page interaction. Use it when you need to scrape web pages, extract structured data, take screenshots, parse PDFs, or crawl entire websites. Triggers: firecrawl, scrape, extract content, screenshot, parse pdf, crawl website, fetch web pages, extract content, webpage screenshot

Access device web pages through a browser to inspect page structure, extract displayed data, and validate live media streams.

DBOS TypeScript SDK for building reliable, fault-tolerant applications with durable workflows. Use this skill when writing TypeScript code with DBOS, creating workflows and steps, using queues, events, messages, and streams.

Creates comprehensive technical documentation from existing codebases. Analyzes architecture, design patterns, and implementation details to produce long-form technical manuals and ebooks.

Deep knowledge for writing, reviewing, and debugging PowerShell scripts on Windows (5.1) and cross-platform (PowerShell Core 7+). Covers error handling, cmdlet conventions, pipelines, modules, security, and common automation idioms. Load when the user is working with .ps1, .psm1, or .psd1 files, or shell automation targeting Windows or PowerShell Core.

Proves or disproves convergence of RL algorithms using contraction mapping, Dvoretzky, Extended Dvoretzky, and Robbins-Monro theorems.

Query database safely with parameterized statements

Check Meilisearch index status, tasks, health, and settings. Use for debugging search issues, monitoring indexing tasks, and inspecting index configuration. Read-only admin operations.

Reserved and on-demand GPU cloud instances for ML training and inference. Use when you need dedicated GPU instances with simple SSH access, persistent filesystems, or high-performance multi-node clusters for large-scale training.

Expert guidance for fine-tuning LLMs with LLaMA-Factory - WebUI no-code, 100+ models, 2/3/4/5/6/8-bit QLoRA, multimodal support

Guide for creating high-quality MCP (Model Context Protocol) servers that enable LLMs to interact with external services through well-designed tools. Use when building MCP servers to integrate external APIs or services, whether in Python (FastMCP) or Node/TypeScript (MCP SDK).

Serverless GPU cloud platform for running ML workloads. Use when you need on-demand GPU access without infrastructure management, deploying ML models as APIs, or running batch jobs with automatic scaling.

High-level PyTorch framework with Trainer class, automatic distributed training (DDP/FSDP/DeepSpeed), callbacks system, and minimal boilerplate. Scales from laptop to supercomputer with the same code. Use when you want clean training loops with built-in best practices.

Guides REST and GraphQL API design, endpoint patterns, request/response schemas, versioning, and API best practices. Use when building APIs, designing endpoints, or reviewing API contracts.

Source-grounded paper reading skill for Codex and Claude Code-style local AI agent workflows. Generates a deep reading report, traceability artifacts, and an optional evidence reader.