Claude Skills
The open format is called Agent Skills and works in Claude Code, Codex, Cursor and other agents — most people know it as Claude Skills.
Every Agent Skill we could find on GitHub, deduplicated by content. 80 149 files from 1 774 authors, of which 62 489 are unique — the rest is the same skill repackaged into someone else's repository. For each one: what it weighs in tokens, whether it ships runnable scripts, and which MCP servers it needs.
> Authoritative guidance for building Claude Code skills, agents, and plugins, plus init and update steps that install and refresh the plugin-building skills in the current repository. Use when you need the rules or best practices for a skill, agent, hook, or plugin — designing, reviewing, hardening, or checking one against the guidance. Run with `init` to vendor the guidance, skill-builder, and agent-builder skills into the current repository (so they run with no dependency on this plugin) plus a path-scoped rule index, or `update` to refresh an already-vendored copy. Does not run an interview to build a new skill or agent from scratch — use skill-builder or agent-builder. Does not write feature code, review application code, or build non-plugin features.
> Builds a new Claude Code skill from scratch through a relentless, evidence-based interview that walks the skill's design tree decision-by-decision — entity fit, use cases, name, description, workflow steps, tools, and progressive-disclosure layout — then reviews the finished skill against the plugin-building guidance and applies every fix it finds. Use when creating, authoring, scaffolding, designing, or drafting a new skill or slash command. Does not build an agent or subagent — use agent-builder. Does not serve, vendor, or refresh the authoring guidance itself — use guidance.
> Convert a stakeholder summary markdown file into a single self-contained HTML executive report — bottom line and decision asks up front, supporting detail later — styled with a Test Double-derived palette and self-contained mermaid diagrams. Use when the user wants to turn a stakeholder summary, executive summary, or business summary into an HTML report, generate an HTML version of a summary doc, or produce a shareable HTML file from a summary markdown. Produces an HTML sibling file only; does not publish anything.
> Authoritative guidance for building Claude Code skills, agents, and plugins, vendored into this repository. Use when you need the rules or best practices for a skill, agent, hook, or plugin — designing, reviewing, hardening, or checking one against the guidance. Does not run an interview to build a new skill or agent from scratch — use skill-builder or agent-builder. Does not write feature code, review application code, or build non-plugin features.
> Produces a plain-language stakeholder summary from an existing feature specification, for sharing with non-technical stakeholders before implementation kicks off. Use when the user wants to draft a stakeholder summary, executive summary, or business summary of a feature spec or PRD. Does not write the spec itself — use plan-a-feature. Does not sequence the build into phases — use plan-a-phased-build. Does not produce an implementation plan — use plan-implementation.
>- 根据策略描述、研报 PDF 或截图，解读因子/选股逻辑，基于 scripts/daily-factors-backtest.py 框架生成 QMT 内置日频因子回测脚本。用户提到 QMT 内置回测、因子选股回测、截面因子、 研报复现、handlebar 回测、after_init 预计算信号时使用。
Guide for creating effective skills that extend agent capabilities with specialized knowledge, workflows, or tool integrations. Use this skill when the user asks to: (1) create a new skill, (2) make a skill, (3) build a skill, (4) set up a skill, (5) initialize a skill, (6) scaffold a skill, (7) update or modify an existing skill, (8) validate a skill, (9) learn about skill structure, (10) understand how skills work, or (11) get guidance on skill design patterns. Trigger on phrases like \"create a skill\", \"new skill\", \"make a skill\", \"skill for X\", \"how do I create a skill\", or \"help me build a skill\".
Query current NVIDIA CUDA, PTX ISA, Runtime API, Driver API, Programming Guide, Best Practices, Nsight Compute, and Nsight Systems references. Use for direct CUDA C++ or PTX work, and for framework tasks only when they need NVIDIA ISA, API, architecture, or tool facts. Triggers include inline PTX, WMMA, WGMMA, TMA, tcgen05, mbarrier, fabric operations, CUDA APIs and Graphs, memory ordering, compute capability, Ampere, Hopper, Blackwell, Rubin, nsys, ncu, and compute-sanitizer.
Write, debug, and optimize CUTLASS and CuTeDSL GPU kernels using local source code, examples, and header references. Use when the user mentions CUTLASS, CuTe, CuTeDSL, cute::Layout, cute::Tensor, TiledMMA, TiledCopy, CollectiveMainloop, CollectiveEpilogue, GEMM kernel, grouped GEMM, sparse GEMM, flash attention CUTLASS, blackwell GEMM, hopper GEMM, FP8 GEMM, FP4 GEMM, blockwise scaling, MoE GEMM, StreamK, warp specialization CUTLASS, TMA CUTLASS, epilogue fusion, EVT (Epilogue Visitor Tree), pycute, Layout algebra, Swizzle pattern, GemmUniversal, KernelSchedule, EpilogueSchedule, CUTLASS collective builder, CUTLASS pipeline, or asks about writing high-performance CUDA kernels with CUTLASS/CuTe templates. Also use when the user wants to understand CUTLASS source code structure, compile CUTLASS examples, or debug CUTLASS template errors.
Develop, debug, and optimize SGLang LLM serving engine. Use when the user mentions SGLang, sglang, srt, sgl-kernel, LLM serving, model inference, KV cache, attention backend, FlashInfer backend, MLA, MoE routing, MoE dispatch, expert parallelism SGLang, speculative decoding, disaggregated serving, TP/PP/EP, radix cache, continuous batching, chunked prefill, CUDA graph SGLang, model loading, quantization FP8/GPTQ/AWQ, JIT kernel, triton kernel SGLang, DeepSeek serving, EPLB (expert load balancing), HiCache, launch_server, sglang Engine API, LoRA inference, torch.compile SGLang, or asks about serving LLMs with SGLang. Also use when the user wants to add a new model to SGLang, add a new attention backend, debug SGLang serving issues, or optimize SGLang throughput/latency.
Write, debug, and optimize Triton and Gluon GPU kernels using local source code, tutorials, and kernel references. Use when the user mentions Triton, Gluon, tl.load, tl.store, tl.dot, tl.dot_scaled, triton.jit, gluon.jit, wgmma, tcgen05, TMA, tensor descriptor, persistent kernel, warp specialization, fused attention, matmul kernel, kernel fusion, tl.program_id, triton autotune, MXFP, FP8, FP4, NVFP4, block-scaled matmul, SwiGLU, top-k, triton_kernels, roofline analysis, Triton IR, TritonGPU dialect, MLIR Triton, PDL (programmatic dependent launch), cluster launch control, or asks about writing GPU kernels in Python. Also use when the user wants to understand Triton compiler internals, debug Triton kernel correctness, profile Triton kernel performance, or convert CUDA kernels to Triton.
> writing, editing, creating documents, reports, articles, READMEs, notes, outlines, research summaries, translations, restructuring, formatting, or any markdown content. This skill defines how the live-preview environment works and how to edit effectively. Consult before your first edit in a new conversation.
> creating or editing architecture diagrams, flowcharts, UML, ER diagrams, network topology, org charts, mind maps, or any draw.io diagram. This skill defines the .drawio XML format, style strings, edge rules, and color palette. Consult before your first edit in a new conversation.
Project-context awareness — multi-session workflows around one topic, shared materials and preferences, and cross-mode handoffs as a high-value user path.
AI-orchestrated video production on @pneuma-craft. Use whenever the user wants to generate, edit, or compose video clips, audio tracks, captions, or background music — including text-to-video / image-to-video generation, TTS narration, music generation, provenance tracking, and timeline composition. Trigger on phrases like "generate video", "make a clip", "add narration", "try another take", "add BGM", "edit project.json", "place on the timeline", "AIGC assets", "regenerate this shot", or any request that touches the exploded timeline or the dive-in panels. Also use when editing `project.json` by hand, registering assets, or wiring provenance edges. Do not assume the user knows the schema — they usually don't; read `references/project-json.md` before committing to an edit.
Author a new Pneuma mode end-to-end — manifest + viewer + skill + seed + showcase. Use this skill whenever the user says they want to create a new mode, fork an existing one for a different domain, scaffold mode files, design a new viewer, or asks "how should I build a mode for X". Walks the user through a discovery interview, produces a design brief that names every key choice (Source kind, ViewerAddress vocabulary, action space, seed strategy, external integrations, evolution directive), and only then generates the directory skeleton. Encodes the practice rules pulled from webcraft / slide / diagram / illustrate / remotion / kami. Pneuma Skills project only; Claude Code only.
> Persistent user preference memory across sessions. Consult this skill BEFORE making any design, style, or aesthetic decisions — choosing colors, themes, layouts, fonts, tone of voice, content density, or visual direction. Also consult when starting a new creative task in any mode, when the user corrects your style choices, or when asked to analyze or refresh user preferences. Even if you think you know what to do, check preferences first — the user may have recorded specific constraints.
> Rewrite the active Pneuma session's UI title + one-line summary so the launcher and ProjectPanel rows reflect what the session is actually about. Use this skill whenever the user asks to "整理 / 概括 / refresh / re-title / summarize this session", whenever the conversation has produced substantive work and the default title ("WebCraft session") is now uninformative, or before the user pauses a long session — the next time they reopen it, the row needs to say reopens, mode-agnostic.
> creating or editing diagrams, flowcharts, wireframes, mind maps, architecture diagrams, org charts, sketches, or any visual content on the Excalidraw canvas. This skill defines the Excalidraw JSON format, element types, binding rules, and color palette. Consult before your first edit in a new conversation.
Paper-canvas web design. Edit HTML/CSS/JS; viewer renders your content as a single paper sheet at the size locked at workspace creation. Design language adapted from tw93/kami (MIT). Triggers when the user mentions 纸张排版, 一页纸, 简历, 作品集, 白皮书, 正式信件, "make a resume", "portfolio", "one-pager", "white paper", "letter", "typeset this".
> Works in a WYSIWYG environment where the user sees edits live in a browser preview panel."
> creating modes, editing manifest.ts, pneuma-mode.ts, viewer components, skill prompts, seed files, publishing, forking, or any mode package development. This skill defines the ModeManifest reference, ViewerContract patterns, and publishing workflow. Consult before your first edit in a new conversation.
> generating images, creating illustrations, editing visuals, managing content sets, organizing rows, crafting prompts, adjusting styles, or any image generation task. This skill defines the generation workflow, manifest format, prompt engineering, and content set organization for the AI illustration studio. Consult before your first edit in a new conversation.
> creating or editing dashboards, adding tiles, changing layouts, updating data sources, adjusting themes, resizing tiles, or any dashboard-building task. This skill defines the defineTile() API, board.json schema, theming conventions, size guidelines, and resize adaptation rules for the live-preview tile grid environment. Consult before your first edit in a new conversation.
A goal-driven Chinese long-form writing partner. Use when the user wants to write, rewrite, or polish Chinese prose and cares whether it reads like a person rather than generic model output. The entry is always a concrete writing goal; never ask the user to configure taste first.
> Pneuma WebCraft Mode workspace guidelines with Impeccable.style design intelligence. styling, animations, responsive design, accessibility, performance optimization, design system extraction, UX writing, and visual refinement. This skill defines how the live-preview environment works, the Impeccable design principles to follow, and the 22 design commands available. Consult before your first edit in a new conversation.
> creating or editing presentations, slide decks, pitch decks, adding or modifying slides, changing themes, layouts, or any presentation content. This skill defines the design workflow, height calculation rules, layout patterns, and quality checklist for the fixed-viewport slide environment. Consult before your first edit in a new conversation.
> {{customDescription}}
Use when mining coding-agent session history, archived transcripts, memories, or repeated local work to discover recurring workflows that should become new Agent Skills.
Use when turning local, private, or personal Agent Skills into publishable skills for GitHub, marketplaces, teams, or public sharing, especially when private paths, personal habits, credentials, internal hosts, or user-specific context must be removed.
Use when auditing or adapting newly created, downloaded, forked, installed, or community Agent Skills to the user's tools, habits, directories, session history, and preferred workflows, especially when triggers feel wrong, noisy, or too generic.
Publish Markdown/HTML articles to WeChat Official Account (微信公众号) drafts via API
Fast synchronous key-value storage for React Native via react-native-mmkv (Nitro-backed). Covers creating and configuring MMKV instances, reading/writing all value types (string, number, boolean, buffer), React hooks for reactive UI, value-change listeners, encryption, AsyncStorage migration, and state-management integrations (zustand, redux-persist, jotai, react-query). Also covers storage limits, multi-process mode, and the V4 upgrade from the class-based API.
Design, implement, and review modern C++ APIs and native implementation code. Use when working on .cpp, .hpp, CMake, RAII ownership, std::variant, callbacks, async work, platform bridges, or C++-backed React Native Nitro Module implementations.
Best-practices guide for react-native-vision-camera v5 (Nitro rewrite, April 2026) and migrating from v4. Use when installing, configuring, or writing code with the Camera, outputs, frame processors, recording, or barcode/depth/RAW features. Also use when converting v4 code (photo={true}, takePhoto, useCameraFormat, useFrameProcessor) to the new v5 API.
Fast native networking primitives for React Native built on Nitro Modules — react-native-nitro-fetch, react-native-nitro-websockets, and react-native-nitro-text-decoder. Covers the fetch API, global replacement, prefetching and cold-start cache warming, the NitroWebSocket class and pre-warming, migrating from React Native's built-in WebSocket, the in-process NetworkInspector, native Perfetto / Instruments tracing, the native TextDecoder, and plugging nitro-fetch into axios via a custom adapter.
Design, implement, and review Swift APIs and Apple-platform code. Use when working on .swift files, Swift types, Foundation or AVFoundation APIs, DispatchQueue, async/await, Task, actors, MainActor, thread-affine state, or Swift-backed React Native Nitro Module implementations.
Builds and designs React Native Nitro Modules with Nitrogen, HybridObject TypeScript specs, Nitro View components, generated native implementations, zero-copy and native-state APIs, Swift/Kotlin/C++ bindings, example apps, and testing. Use when creating a Nitro Module, adding or reviewing HybridObjects, building a Nitro View (HybridView) component, designing Nitro-specific public APIs, implementing native functionality, or setting up the nitrogen codegen pipeline. Pair with api-design for general library API shape.
Design, implement, and review Kotlin and Android APIs. Use when working on .kt files, Kotlin nullability, sealed classes/interfaces, coroutines, Android threading, Java interop, or Kotlin-backed React Native Nitro Module implementations.
Design and review predictable public APIs for TypeScript, JavaScript, React, and React Native libraries. Use when shaping exported functions, classes, hooks, options objects, event and listener APIs, error behavior, naming, cross-platform abstractions, or JS-only packages. Pair with build-nitro-modules when the library is backed by Nitro.
面向“学习任何知识”的通用导师型 skill。用于用户想学习新主题、制定学习计划、整理学习笔记、做课后复盘、生成项目任务书、做掌握度检查，或希望基于自己资料与外部权威资料进行中文导师式陪学时。默认采用“导师 + 项目教练”模式，以项目驱动和 Mastery Learning 推进；当用户未提供资料时，主动查找官方文档、原始资料、经典教材、权威机构材料与最佳实践，并区分事实依据与建议判断。
> branded PowerPoint. Edits only the slides/objects the user names, using native editable objects, while keeping every other slide, the logo, masters, links, grouping and transparency byte-for-byte unchanged, and writes back a .pptx that opens clean in WPS / PowerPoint. Use automatically when the user asks to revise, fix, refine, restyle, or repair an existing presentation, including phrases such as "改PPT", "修改PPT", "修改这份PPT", "把这个ppt改一下", "调整这几页", "只改第X页", "保真修改", "别动其他页", "edit this pptx", "fix my slides", "revise this deck", "refine these slides", "repair the pptx". It can also generate a new editable PPTX or magazine web deck from source material (PDF/DOCX/XLSX/URL/Markdown) when asked — "create PPT", "make a deck", "生成PPT", "做PPT", "制作演示文稿", "把这个做成PPT", "杂志风PPT", "网页PPT" — but preserve-and-edit is the primary, differentiated mode. Also triggers on "ultimate-ppt-master", "保真改PPT", "deckweaver", or "ppt-master".
Use when a task involves continuing a project, restoring project context, maintaining file-based project memory, updating current-state summaries, or recording meaningful progress across sessions. Works best for long-horizon, file-based projects and supports research writing, product document collaboration, software project coordination, and broader cross-functional project continuity.
> (1) Validating designed sequences fold correctly, (2) Predicting binder-target complex structures, (3) Calculating confidence metrics (pLDDT, pTM, ipTM), (4) Self-consistency validation of designs, (5) Multi-chain complex prediction with AlphaFold-Multimer. For faster single-chain prediction, use esm. For QC thresholds, use protein-qc.
> (1) Designing protein binders with built-in AF2 validation, (2) Running production-quality binder campaigns, (3) Using different design protocols (fast, default, slow), (4) Need joint backbone and sequence optimization, (5) Want high experimental success rate. For backbone-only generation, use rfdiffusion. For QC thresholds, use protein-qc. For tool selection guidance, use binder-design.
> (1) Need side-chain aware design from the start, (2) Designing around small molecules or ligands, (3) Want all-atom diffusion (not just backbone), (4) Require precise binding geometries, (5) Using YAML-based configuration. For backbone-only generation, use rfdiffusion. For sequence-only design, use proteinmpnn. For structure validation, use boltz.
> Goal-oriented binder design campaign planning and health assessment. (2) Converting high-level goals into runnable pipelines, (3) Assessing campaign health and pass rates, (4) Diagnosing why designs are failing QC, (5) Estimating time, cost, and expected yields, (6) Selecting between design tools for a specific target. This skill orchestrates the other protein design tools. For individual tool parameters, use the specific tool skills.
> Guidance for choosing the right protein binder design tool. (2) Planning a binder design campaign, (3) Understanding trade-offs between different approaches, (4) Selecting tools for specific target types. For specific tool parameters, use the individual tool skills (boltzgen, bindcraft, rfdiffusion, etc.).
> (1) Planning binding kinetics experiments, (2) Troubleshooting poor/no binding signal, (3) Interpreting kinetic data artifacts, (4) Choosing between SPR vs BLI platforms.
> Structure prediction using Boltz-1/Boltz-2, an open biomolecular structure predictor. (2) Validating designed binders, (3) Need open-source alternative to AF2, (4) Predicting protein-ligand complexes, (5) Using local GPU resources. For QC thresholds, use protein-qc. For AlphaFold2 prediction, use alphafold. For Chai prediction, use chai.
> ESM protein language models for embeddings, sequence scoring, structure pseudo-log-likelihood (PLL) or mutation-effect scores, (2) Getting protein embeddings for clustering or filtering, (3) Predicting complex structures with ESMFold2, (4) Designing binders by inverting ESMFold2, (5) Filtering designs by sequence plausibility. For diffusion-based structure prediction, use boltz or chai. For QC thresholds, use protein-qc. For gradient-based multi-objective design, use mosaic.
> Multi-objective, gradient-based protein binder design with Mosaic. Use this objective, (2) Optimizing binders against a custom loss rather than a fixed pipeline, (3) Wanting gradient descent over sequence space in the style of ColabDesign, RSO, or BindCraft but with interchangeable predictors, (4) Letting the optimizer choose the epitope instead of fixing hotspots. For an end-to-end binder pipeline with default filters, use bindcraft. For all-atom diffusion design, use boltzgen. For backbone-only generation, use rfdiffusion.
> (1) Finding similar structures in PDB/AFDB databases, (2) Structural homology search, (3) Database queries by 3D structure, (4) Finding remote homologs not detected by sequence, (5) Clustering structures by similarity. For sequence similarity, use uniprot BLAST. For structure prediction, use chai or boltz.
> (1) Designing epitope-targeted nanobodies or scFvs, (2) Needing CDR design on a fixed framework, (3) Working on antibody-format binders rather than miniproteins. For miniprotein binders, use binder-design (BoltzGen, BindCraft, RFdiffusion, Mosaic). For structure validation, use boltz or chai.
Claude Skills — questions
Answers built from the skills we actually parsed.