Provides guidance for training LLMs with reinforcement learning using verl (Volcano Engine RL). Use when implementing RLHF, GRPO, PPO, or other RL algorithms for LLM post-training at scale with...
Evaluates LLMs across 60+ academic benchmarks (MMLU, HumanEval, GSM8K, TruthfulQA, HellaSwag). Use when benchmarking model quality, comparing models, reporting academic results, or tracking...
Extract structured data from LLM responses with Pydantic validation, retry failed extractions automatically, parse complex JSON with type safety, and stream partial results with Instructor -...
Jeffrey Emanuel's multi-agent implementation workflow using NTM, Agent Mail, Beads, and BV. The execution phase that follows planning and bead creation. Includes exact prompts used.
Conducts comprehensive code quality reviews including code smells detection, maintainability assessment, complexity analysis, design pattern evaluation, naming conventions, code duplication,...
Simplify and refine recently modified code for clarity and consistency. Use after writing code to improve readability without changing functionality.
Reduce LLM size and accelerate inference using pruning techniques like Wanda and SparseGPT. Use when compressing models without retraining, achieving 50% sparsity with minimal accuracy loss, or...
Generates well-structured git commit messages following conventional commit standards and best practices. Creates clear, descriptive commits with proper type prefixes (feat, fix, docs, refactor,...
Evaluates LLMs across 100+ benchmarks from 18+ harnesses (MMLU, HumanEval, GSM8K, safety, VLM) with multi-backend execution. Use when needing scalable evaluation on local Docker, Slurm HPC, or...
Conducts comprehensive security code reviews including vulnerability detection (OWASP Top 10, CWE), authentication/authorization flaws, injection attacks, cryptography issues, sensitive data...
Control Ghostty terminal emulator via CLI. Manage windows, tabs, splits, and configuration.
Iterative UI/UX polishing workflow for web applications. The exact prompt and methodology for achieving Stripe-level visual polish through multiple passes.
Multi-cloud orchestration for ML workloads with automatic cost optimization. Use when you need to run training or batch jobs across multiple clouds, leverage spot instances with auto-recovery, or...
Use Bun instead of Node.js, npm, pnpm, or vite. Provides command mappings, Bun-specific APIs, and development patterns.
Replace with description of the skill and when Claude should use it.
Guides comprehensive software project planning including task breakdown, estimation, sprint planning, backlog management, resource allocation, milestone tracking, and risk management. Creates user...
Minimal Chrome DevTools Protocol tools for browser automation and scraping. Use when you need to start Chrome, navigate pages, execute JavaScript, take screenshots, or interactively pick DOM elements.
Guides systematic code refactoring to improve code quality, maintainability, and design. Identifies code smells, applies refactoring patterns, ensures test coverage, and follows safe refactoring...
Provides comprehensive Alibaba Cloud (Aliyun) guidance including ECS, ApsaraDB, OSS, SLB, VPC, RAM, ACK (Kubernetes), Function Compute, API Gateway, CDN, and monitoring services. Covers...
PyTorch library for audio generation including text-to-music (MusicGen) and text-to-sound (AudioGen). Use when you need to generate music from text descriptions, create sound effects, or perform...