所有 Skills
找到 8065 个 Skills
Skills 列表
aoti-debug
排查与调试 AOTInductor (AOTI) 的错误与崩溃问题。当遇到 AOTI 段错误（segfault）、设备不匹配错误、常量加载失败，或者调用 aot_compile、aot_load、aoti_compile_and_package、aoti_load_package 触发运行时错误时使用。
pytorchadd-uint-support
通过更新 AT_DISPATCH 宏，为 PyTorch 算子添加无符号整数（uint）类型支持。适用于为算子或 Kernel 补充 uint16、uint32、uint64 类型支持的场景，或者当用户提到开启无符号类型支持、基础无符号类型（barebones unsigned types）或 uint 支持时使用。
pytorchat-dispatch-v2
在 ATen C++ 代码中将 PyTorch 的旧版 AT_DISPATCH 宏转换为全新的 AT_DISPATCH_V2 格式。适用于将 AT_DISPATCH_ALL_TYPES_AND*、AT_DISPATCH_FLOATING_TYPES* 或其他分发宏迁移至新版 v2 API 的场景。主要用于 ATen 算子内核文件、CUDA kernel 以及原生算子（native operator）实现。
pytorchmetal-kernel
为 PyTorch 算子编写 Metal/MPS kernel。适用于给算子添加 MPS 设备支持、实现 Metal 着色器或将 CUDA kernel 移植到 Apple Silicon 的场景。涵盖 native_functions.yaml 分发配置、主机端算子实现以及 Metal kernel 的编写。
pytorchpyrefly-type-coverage
将指定文件迁移至更严格的 Pyrefly 类型检查模式，要求所有函数、类以及属性必须补全类型标注。
pytorchtriaging-issues
对 GitHub Issue 进行分类分流（Triage）：自动路由至 Oncall 值班团队、打上合适 Label，以及自动回复并关闭纯咨询类提问。适用于处理 PyTorch 新提交的 Issue，或收到 Issue 分流处理指令时使用。
pytorchpr-review
审查 PyTorch 的 Pull Request（PR），重点检查代码质量、测试覆盖率、安全性以及向下兼容性（BC）。适用于 PR 审查、代码变更评审，或者当用户提及“review PR”、“code review”、“帮我看下这个 PR”等场景。
pytorchdocstring
按照 PyTorch 官方规范为 PyTorch 函数和方法编写 docstring（文档字符串）。在编写或更新 PyTorch 代码中的 docstring 时使用。
pytorchskill-writer
引导用户为 Claude Code 创建 Agent Skill。当用户想要创建、撰写、设计新 Skill，或者在 SKILL.md 文件、frontmatter 配置、Skill 目录结构上需要帮助时使用。
pytorch
fastapi
FastAPI 最佳实践与编码规范。适用于开发 FastAPI API、定义 Pydantic 模型、处理依赖注入、实现流式响应（含 SSE 服务端推送）以及托管前端应用。帮助你保持 FastAPI 代码整洁优雅，并紧跟最新的特性与设计模式。
fastapi
reference-signal-forms
Explains the mental model and architecture of the code under `packages/forms/signals`. You MUST use this skill any time you plan to work with code in `packages/forms/signals`
angularangular-developer
生成 Angular 代码并提供架构指导。在创建项目、组件或服务时，或针对响应式（信号、linkedSignal、resource）、表单、依赖注入、路由、SSR、无障碍（ARIA）、动画、样式（组件样式、Tailwind CSS）、测试或 CLI 工具的最佳实践时触发。
angular
browser-use
自动化浏览器交互，用于网页测试、表单填写、截图和数据提取。当用户需要浏览网站、与网页交互、填写表单、截图或从网页提取信息时使用。
browser-use
caveman-explore
Read-only repository explorer. Use PROACTIVELY for cold-start exploration, broad cross-file localization, or when a direct search has failed and you need to find where something lives. Skip it when the issue already names the exact file or symbol, or a previous turn already returned usable file:line evidence. Returns only compact path:line citations; its reads and greps never enter the main conversation.
juliusbrusseecaveman-optimize
Turn Caveman's exact report-only repository observations into an operator-chosen optimization candidate with a paired baseline/candidate evaluation. Use when the user asks to inspect an optimization observation, evaluate a candidate change, or act on the current Caveman optimization report. Require a logged-in Caveman CLI connection and explicit approval; never infer money or actuation from a profile.
juliusbrusseeinvestigate-first
Diagnose ambiguous failures before editing. Use for unknown causes, intermittent behavior, performance regressions, or investigations needing evidence-ranked hypotheses.
juliusbrusseeverify-and-stop
Prove existing work meets acceptance conditions without expanding scope. Use for validation-only tasks, completion checks, focused gate runs, and last-mile proof.
juliusbrusseesurgical-patch
Fix bugs and small behavior changes at the narrowest responsible layer. Use when regression proof, preserved surrounding behavior, and task-relevant tests matter.
juliusbrusseesafe-refactor
Restructure code while preserving behavior. Use for extraction, consolidation, ownership moves, or cleanup where verification must bracket structural edits.
juliusbrusseecaveman-learn
Close the loop on a Caveman learn report — review the ranked token sinks and apply cost-lowering fixes (trim config, offload recurring context to cavemem) with per-edit consent. Use when the user runs "caveman learn", asks to lower their agent's token cost, wants to trim a heavy CLAUDE.md, or wants to offload context they re-paste every session into cavemem.
juliusbrusseelean-build
Build feature work with high overbuilding risk. Use for new behavior, product slices, or integrations where repository reuse, strict scope, and an explicit stop condition matter.
juliusbrusseecaveman-evidence-review
Review Caveman Cloud evidence read-only: costs, Cave Score, Cave Plan, workflows, traces, latency, errors, compression, routing, and verified savings. Use when the user asks what Caveman found, where LLM spend goes, why cost or quality changed, which workflows need attention, or asks for a trace or analytics review. Prefer Caveman MCP tools; fall back to CLI JSON.
juliusbrusseecaveman-discover
Find every LLM workflow in the current repository and label it, so Caveman Cloud groups spend by what the code actually does (support-reply, nightly-digest) instead of one anonymous bucket. Use when the user pastes the Caveman discovery prompt, says "discover workflows", or asks to break LLM spend down by workflow. The repo should already route through the Caveman gateway (the caveman-setup skill does that part).
juliusbrusseecaveman-manage
Inspect Caveman Cloud's eval-gated experiment lifecycle and block unsafe execution. Use when the user asks to start, approve, cancel, promote, or roll back a Caveman experiment, or asks what action an experiment's evidence supports. Read evidence first; do not execute lifecycle mutations until server-authoritative transition and evidence gates ship.
juliusbrussee