583 skills · sorted by relevance
Make system architecture decisions by clarifying requirements and constraints, comparing benefits and drawbacks, and considering simpler alternatives. Record the reasoning in ADRs, architecture decision records, so significant choices and their rationale are documented.
Evaluates task complexity and selects a small set of existing skills for work spanning multiple areas. Uses agent-memory-mcp to record and retrieve successful combinations, while keeping simple tasks within the agent’s ordinary capabilities.
Create, edit, inspect, and verify PowerPoint slide decks in PPTX format. Use a consistent, simple layout and consolidate information into fewer slides, then reopen the presentation to check its slide count and titles.
Test whether you understand a concept and whether evidence supports a decision using frameworks drawn from Richard Feynman’s work. Explore concrete examples, simple demonstrations, counterevidence, and the limits of analogies to expose gaps in reasoning.
Train language models with SimPO using pairs of preferred and rejected answers. Configure preference training without a separate reference model, adapt settings for base or instruction models, and address unstable training, memory limits, or lost capabilities.
Adapt PyTorch training scripts for multiple GPUs or machines with Accelerate. This skill guides code changes, configuration, and launch commands, including mixed precision, which combines numerical formats, and DeepSpeed or FSDP integration for distributed training.
Manage separate browser profiles for website tasks, keeping each account’s cookies and logins between sessions. The skill guides profile selection, reuse, opening, and closing, and explains when a direct download is simpler than browser work.
Turn rough bug reports, feature requests, support notes, or pull requests into concise issues. Explain the current problem and desired behavior in plain language, preserving useful evidence without prescribing how developers should implement a solution.
Keep coding tasks small by checking whether new code is needed, reusing existing helpers, and preferring standard libraries or built-in platform features. Apply the same approach to feature development, bug fixes, refactoring, and reviews.
Create a small interactive HTML playground to explore how something works. It packages an explanation, simulation, or simple game into one screen, with clearly labeled keyboard controls and no network requests or external assets.
Guide language model training with SimPO using pairs of preferred and rejected answers. Covers configuration, learning rate tuning, and memory troubleshooting for preference alignment, which adjusts model responses toward human preferences without a separate reference model.
Adapt existing PyTorch training scripts to run across multiple GPUs or machines with Accelerate. This skill covers device setup, distributed training configuration, and mixed precision, which uses lower-precision numbers to reduce memory use.
Convert Markdown into Word, PDF, and HTML using a workflow suited to the document. Simple files use direct delegation, while documents with special characters use separate conversion steps, character replacement, and output checks.
Add compact visual summaries to Markdown answers using static HTML and CSS. Show comparisons, key metrics, timelines, or simple relationships in a preview, while keeping explanations and detailed information in the surrounding text.
Create simple, cute square character images using rounded shapes, two character colors, and a solid background. The skill can use product context to propose three design directions and generate six separate candidates for comparison.
Control deep thinking mode. Default enabled. For very simple tasks (simple reminders, greetings, quick queries), can temporarily disable to speed up response. Auto-restores to enabled after completion.
The English summary is being prepared. Showing the original description.
Generate game content procedurally — seeded deterministic RNG, value/Perlin/ Simplex noise for terrain and heightmaps, grid dungeon generation (rooms + corridors, BSP, random walk), and weighted loot/drop tables. Engine-neutral algorithms. Use when the user mentions procedural generation, perlin/simplex noise, random seed, dungeon generator, heightmap/terrain, or loot tables.
The English summary is being prepared. Showing the original description.
ALWAYS invoke this skill when the user asks for simpler or shorter about something said or written - "say it simply", "what does this mean", "I don't understand your answer", "too long", "wait, what?", "bro" - in any language, about any text: your own answer, a report, a review comment, an error. "I don't understand what to DO" is os-step-by-step; this skill restates text. It restates for a reader who does not read code: leads with the point, keeps every number, warning and caveat - no facts added, no bad news dropped. A number returns exactly that many points, most important first.
The English summary is being prepared. Showing the original description.
Simplest distributed training API. 4 lines to add distributed support to any PyTorch script. Unified API for DeepSpeed/FSDP/Megatron/DDP. Automatic device placement, mixed precision (FP16/BF16/FP8). Interactive config, single launch command. HuggingFace ecosystem standard.
The English summary is being prepared. Showing the original description.
Iteratively study and apply elegant coding patterns. Each iteration - understand the code, research what simple and elegant code looks like, apply learnings, verify with CI. Use as a standalone refactoring pass or when the user asks to make code more elegant, simple, or idiomatic.
The English summary is being prepared. Showing the original description.
Register, segment, filter, resample 3D medical images (MRI, CT, microscopy) via SimpleITK Python; DICOM, NIfTI, multi-modal. Rigid/affine/deformable registration, threshold/region-growing segmentation, Gaussian/morph filtering, label stats, format conversion. Use to align volumes across timepoints/modalities, segment fluorescence, or convert DICOM→NIfTI.
The English summary is being prepared. Showing the original description.
通用 horizontal-swipe HTML deck, 不要 magazine 调
The English summary is being prepared. Showing the original description.
Iteratively study and apply elegant coding patterns. Each iteration - understand the code, research what simple and elegant code looks like, apply learnings, verify with CI. Use as a standalone refactoring pass or when the user asks to make code more elegant, simple, or idiomatic.
The English summary is being prepared. Showing the original description.
GraphQL gives clients exactly the data they need - no more, no less. One endpoint, typed schema, introspection. But the flexibility that makes it powerful also makes it dangerous. Without proper controls, clients can craft queries that bring down your server. This skill covers schema design, resolvers, DataLoader for N+1 prevention, federation for microservices, and client integration with Apollo/urql. Key insight: GraphQL is a contract. The schema is the API documentation. Design it carefully. 2025 lesson: GraphQL isn't always the answer. For simple CRUD, REST is simpler. For high-performance public APIs, REST with caching wins. Use GraphQL when you have complex data relationships and diverse client needs. Use when "graphql, graphql schema, graphql resolver, apollo server, apollo client, graphql federation, dataloader, graphql codegen, graphql query, graphql mutation, graphql, api, apollo, schema, resolvers, dataloader, federation, typescript" mentioned.
The English summary is being prepared. Showing the original description.