Coding standards, frameworks, refactoring, and code review (16.8K skills)
Add a diffusion model or Diffusers generation pipeline to SGLang. Examine reference code, choose a suitable pipeline structure, reuse existing components, and implement model settings and processing stages, with guidance for later multi-GPU support.
Apply consistent names to SGLang speculative decoding code, which verifies draft tokens to generate text. Distinguish tokens with or without the bonus token, counts, counters, and arrays when adding identifiers, renaming code, or reviewing changes.
Guide changes to SGLang runtime configuration and shared process state. Identify where resolved settings, runtime flags, resources, and per-forward state belong, and use the documented access patterns and checks when developing or testing.
Add a CUDA/C++ GPU operation to sgl-kernel using AOT, meaning compilation before use. Follow the implementation through PyTorch registration, Python API exposure, SGLang integration, correctness tests, and benchmarks that account for cache reuse.
Investigate CUDA crashes in SGLang using logs and inputs saved before execution. Inspect tensor shapes, data types, and numerical anomalies, with guidance for separate process logs and additional CUDA debugging tools.
Add, rename, or review SGLang environment variables using consistent definitions, types, names, and access methods. The skill also guides migration from old names and temporary test overrides that restore the original values afterward.
Investigate Spring configuration and Beans, the objects Spring manages, using Arthas. Check configuration sources, find registered objects by name or type, and inspect registration details while prioritizing read-only queries.
Investigate high CPU usage in Java applications with Arthas. Identify busy threads, inspect method call paths, and produce an evidence-based diagnosis with specific methods to examine next.
Diagnose running Java applications with Arthas. Follow guides for high CPU usage, retrieving EagleEye request tracking IDs, and inspecting Spring configuration and managed objects, then summarize evidence and next steps.
Rebuild existing web pages as working, editable local code using their actual content and assets. Inspect desktop and mobile layouts, reproduce interactions and animations, then compare rendered pages and fix visible or functional differences.
Convert a Vibe-Trading backtest strategy or a written trading strategy into a Python class for vnpy. Map indicators and trading signals to vnpy conventions, with asset-specific order handling and instructions for loading the strategy for backtesting or live trading.
Translate a Python trading strategy or written trading rules into code for TradingView, 通达信, 同花顺, 东方财富, and MT5. Receive platform-specific indicators or strategy code, setup instructions, and notes explaining features that cannot be translated directly.
Write signal generation code for backtests combining stocks, crypto, forex, and other markets. Apply market-specific indicator settings and volatility-adjusted weights, while the existing CompositeEngine handles trading calendars, shared capital, and each market's rules.
Classify skills CLI pull requests, or proposed code changes, as documentation, bug fixes, or enhancements using their descriptions and patches. Add one label when the evidence is clear, while preserving existing classification labels.
Assess skills CLI pull requests for focused bug fixes using visible patches and regression tests. Surface well-supported candidates with a label and brief maintainer comment, while keeping uncertain findings out of public updates.
Inspect, run, and organize project commands with just. This skill explains how to find available recipes—named tasks in a justfile—check their contents, pass arguments, and define reusable commands with dependencies.
Configure how the browser starts, including window visibility, proxy settings, extra launch options, and a custom browser executable. It also explains opening a local debugging port so other local tools can connect.
Turn reusable decisions, templates, and workflows from a conversation into a workspace skill. It guides planning, user approval, drafting, validation, and publication, with optional behavior tests to check how the resulting skill performs.
Find relevant QwenPaw documentation and source files for questions about setup, configuration, skills, memory, and other features. It maps question keywords to likely file paths so the agent can read and verify the right material.
Configure tools and skills for sub-agents before dispatching development tasks. Role presets cover implementation, planning, reviews, testing, and verification, with defined boundaries for what each agent should do and whether it may modify files.
Find the QwenPaw documentation and source files most relevant to a question. This skill maps topics such as installation, configuration, skills, and memory to likely reading paths, helping an agent focus its investigation.
Configure browser startup with a visible or hidden window, and optionally open a local debugging port for another tool to connect. The skill explains launch parameters, compatibility behavior, and when a browser restart is required.
Connect to an existing Chrome browser or launch one with an explicit debugging port for sharing across agents and tools. It also covers scanning local CDP ports, the connection endpoints used to control a browser.
Integrate a model or pipeline into diffusers, from file structure and weight conversion to tests and documentation. Verify numerical agreement with the original implementation, with modular pipelines as the preferred starting point for new integrations.