Loading...
Loading...
Found 1,268 Skills
State-of-the-art Machine Learning for PyTorch, TensorFlow, and JAX. Provides thousands of pretrained models to perform tasks on different modalities such as text, vision, and audio. The industry standard for Large Language Models (LLMs) and foundation models in science.
Chain-of-Verification (CoVe) prompting system. Converts lazy prompts into rigorous 4-stage verified output. Use for any code generation, debugging, or implementation task. Automatically invoked by wavybaby for medium/high complexity tasks. Reduces hallucinations and catches subtle bugs.
Use when user wants to find a note to publish as a blog post. Triggers on「选一篇笔记发博客」「note to blog」「写博客」「博客选题」. Scans Obsidian notes via Python script, evaluates blog-readiness, supports batch selection with fast/deep dual-track and parallel Agent dispatch.
Form a high-level investment committee consisting of three virtual experts modeled after legendary investors (Buffett, Wood, Druckenmiller) to conduct independent multi-round adversarial debates. True independent thinking is achieved through physically isolated Gemini API calls, and final resolutions are formed via voting. Use when evaluating investment decisions, reviewing stock research reports, or seeking multi-perspective analysis on public companies.
Configure LangChain local development workflow with hot reload and testing. Use when setting up development environment, configuring test fixtures, or establishing a rapid iteration workflow for LangChain apps. Trigger with phrases like "langchain dev setup", "langchain local development", "langchain testing", "langchain development workflow".
Creates and reviews CLAUDE.md configuration files for Claude Code. Applies HumanLayer guidelines including instruction budgets (~50 user-level, ~100 project-level), WHAT/WHY/HOW framework, and progressive disclosure. Identifies anti-patterns like using Claude as a linter for style rules.
Turn a vague research direction into a problem-anchored, elegant, frontier-aware, implementation-oriented method plan via iterative GPT-5.4 review. Use when the user says "refine my approach", "帮我细化方案", "decompose this problem", "打磨idea", "refine research plan", "细化研究方案", or wants a concrete research method that stays simple, focused, and top-venue ready instead of a vague or overbuilt idea.
Find prompt and model quality issues using real conversation data, with specific optimization recommendations. Can implement prompt fixes and model switches directly in your codebase.
Read and analyze arXiv papers by fetching LaTeX source, listing sections, or extracting abstracts. Use when the user mentions arXiv, research papers, preprints, paper IDs like 2301.xxxxx, or wants to read academic publications.
Fact-forcing gate that blocks Edit/Write/Bash (including MultiEdit) and demands concrete investigation (importers, data schemas, user instruction) before allowing the action. Measurably improves output quality by +2.25 points vs ungated agents.
Cost-conscious Claude Code mode. Reduces output tokens 40-70% and overall costs 30-60% by enforcing concise responses, smart model routing, and efficient workflow patterns. Keeps full technical accuracy. Activate with /cost-mode or "enable cost mode". Auto-triggers on mentions of budget, cost, tokens, or spending.
Causal digital twin for marketing simulation — predict campaign ROI, run counterfactual KOL swaps, and audit causal graphs before spending a dollar.