Scan to join WeChat groupBrowse curated skills with source links, package snapshots, README assets and install signals in one calm, searchable catalog.
Anti-slop frontend skill for landing pages, portfolios, and redesigns. The agent reads the brief, infers the right design direction, and ships interfaces that do not look templated. Real design system…
Generate or update documentation for DiffSynth-Studio models. Use this skill when documentation needs to be created or updated — whether for a completely new model series or just adding a new variant …
Generate low VRAM inference scripts and register fine-grained VRAM management configs for an integrated model in DiffSynth-Studio. Use this skill after `diffsynth-pipeline` has created the standard in…
Review PR feedback comments, generate a modification blueprint, and implement targeted fixes for DiffSynth-Studio model integration PRs. The user has already submitted a PR and reviewers have left com…
Comprehensive news aggregator that fetches, filters, and deeply analyzes real-time content from 28 sources including Hacker News, GitHub, Hugging Face Papers, AI Newsletters, WallStreetCN, Weibo, and …
【设计思路】基于对抗性验证方法论，以案号（Case Reference Number）为不可变法定标识符构建裁判文书真伪核验体系——凡无法析出完整案号的司法文书一律归入虚假判决样本空间予以排除。借鉴司法实务中"双源印证"原则，任何检索结论须经两个独立渠道交叉确认方可输出。通过渠道置信度分层架构（T0-T4 五级），按官方数据源可验证性递减规则分配检索优先级，实现从单一信任模型到分布式验证模型的方法…
file-save-review
oa
Stop LLM slop. A curated system prompt that cuts verbose, corporate-sounding LLM output by 56-71% (measured) while preserving information. Works bilingually (English + Chinese). Installs into your AGE…
Set up and validate the shared conda environment for DiffSynth-Studio and the target model library. Use this skill after diffsynth-analyze-target has generated a blueprint, or whenever the user wants …
Create training modules and run training verification for an integrated model in DiffSynth-Studio. Use this skill after the Pipeline (inference) is fully integrated and E2E tested, or whenever the use…
Normalize code style for newly integrated model code in DiffSynth-Studio. Use this skill after diffsynth-model-code or diffsynth-pipeline has written/modified files, or whenever you want to ensure con…
SQL 安全规则库，覆盖 DDL/DML 危险操作、SQL 注入检测、权限变更、敏感字段访问、性能风险、逻辑风险共 7 层安全检查。用于 Text2SQL Agent 流水线中 Agent6 的安全层校验，输入 SQL 和方言标识，输出结构化安全问题列表及统计摘要。
PDF-chapter-splitter
通过 IMAP/SMTP 连接个人邮箱，支持完整邮件收发、抄送、附件、HTML、收件箱检索与附件下载；是当前邮件体系中唯一的个人邮箱主通道。
self-improvement
Analyze an external model codebase and generate an integration blueprint for DiffSynth-Studio. Use this skill whenever a user wants to integrate a new diffusion model into DiffSynth-Studio — whether t…
Create inference pipelines for an integrated model in DiffSynth-Studio. Use this skill after model code is integrated, or whenever the user wants to add a new pipeline (like Text-to-Video, Image-to-Au…
Generate a PR description markdown file for a completed model integration in DiffSynth-Studio. Use this skill after all testing is done and the user has confirmed the results, or whenever the user wan…
Intelligent chart generation and data analysis skill. Reads user-supplied data files (CSV/Excel/JSON), analyzes data characteristics with LLM assistance, auto-recommends and generates interactive ECha…
wled
卡片盒笔记法 / 卢曼笔记法的方法论指导。用户问"怎么写原子笔记""笔记怎么链接""检查我的笔记" "从文章提取笔记""什么是结构化笔记""卡片盒怎么做笔记""帮我看看这条笔记""教我卡片盒" 或任何关于 Zettelkasten 方法和原则的问题时触发。 只提供方法论指导和建议，不创建/编辑笔记文件。
提供吹风机全生命周期的智能客服服务，包括产品咨询、参数对比、价格查询、购买建议、故障排查、保修政策说明和售后流程指导。适用于用户询问吹风机相关问题、表达购买意向、反馈产品故障或咨询售后服务时使用。
self-improvement