AI & ML

saarthurr Bundle

Autonomous orchestrator for serious, codebase-wide work. Use whenever the user says "swarm" or hands over a broad task like "fix the bugs", "figure out why X isn't working", "make this production-ready", "audit/harden this", or "go over the whole codebase". Dispatches many sub-agents in parallel to explore, diagnose, fix, and review; cross-checks every conclusion with independent sub-agents to avoid hallucination; keeps its own context lean by having sub-agents return only distilled findings; survives compaction via a durable ledger and reuses per-repo learnings across runs; pulls full context from every named tool, MCP, and plugin (Supabase, Render, Vercel, Slack, etc.); writes code via the ponytail skill under hard safety guardrails; and does NOT stop until the result is verified production-ready.

sakurakilove Bundle

连接 Facetory 内置 MCP 服务器，检查、创建、编辑、验证、保存和导出小米表盘项目，涵盖普通与息屏主题、动态数据绑定、本地图片/字体导入、图层顺序、可复用模板和导出排错。用于通过 facetory:// 资源或 facetory.* MCP 工具操作 Facetory、诊断内置服务器、续作已打开的 .facetory 项目、制作 Xiaomi/Redmi/Mi Band 表盘、修复 AOD 或导出错误，以及整理成品资源。

salmonbird Bundle

Use Evoui Browser as the default entry point for common, self-terminating web tasks that need a real browser through agent-browser, including navigation, page reading, clicks, forms, login flows, screenshots, web testing, and browser automation. Managed tasks record evidence from the first run for possible reuse. Also use it when the user explicitly asks to execute a complete agent-browser command unchanged. Unless the user requests such an exact command, do not use it for semantic operations that already have a dedicated connector or API, tasks that only require the host's existing Browser or Chrome state, desktop applications, or work that only needs an API or local files. Route interactive or long-running agent-browser invocations through Transparent.

samuele95 Bundle

Build-time meta-prompting framework that designs, audits, and refactors prompts for tool-calling LLM agents. Use whenever the user wants to write, design, build, improve, refactor, or audit any prompt — a system prompt, agentic-loop seed brief, one-shot task, workflow chain, sub-agent or tool description, system persona, LLM-as-judge or grader rubric, or multi-agent team. Trigger on requests like "write a prompt for X", "I need a system prompt", "design a grader", "build an agent", "set up a Claude Code prompt", "fix/improve/refactor my prompt", or when the user invokes /prometheus — and even when the user never says "prompt" but is clearly trying to get a model to behave a certain way. Produces the prompt in the right structural shape, audits and scores it honestly, and ships a runnable verifier plus cognitive-tool scaffolds for hard-reasoning tasks. Not for simple Q&A or when the user wants a direct answer rather than a reusable prompt.

scarxparth Bundle

Audits any project for security, test coverage, and Claude Code automation. 46 checks across 6 layers: secrets, CLAUDE.md, hooks, skills, agents, rules, MCP, CI, memory. Adaptive scoring — auto-detects project maturity (Starter/Growing/Mature/Pro). Diagnoses gaps, prescribes project-specific fixes, applies them. Usage: /doctor — full audit (all 6 layers) /doctor quick — TOP-3 priorities + score (30 seconds) /doctor scan — diagnose only (no changes) /doctor fix — prescribe + apply fixes /doctor layer <N> — audit specific layer (0-5) /doctor verify — health check Triggers: "/doctor", "audit", "setup automation", "configure Claude Code", "project health", "what's missing", "security check", "аудит", "проверь проект"

scottcjn Bundle

Post a swap job over Gensyn AXL, settle onchain via KeeperHub + Uniswap, persist receipt on 0G Storage, return both proofs. Use when an OpenClaw agent needs to ask another agent to perform a paid onchain action and get a verifiable, audit-traceable receipt back. Triggers on intents like "swap X USDC for Y on Base", "post a paid job to peer", "verify a KeeperLink receipt".

sonicbotman Bundle

LobsterPress 认知记忆系统 - DAG 压缩 + OpenClaw ContextEngine + MCP 服务器，让 AI 拥有长期记忆和智能压缩能力

strategy11 Bundle

Create, query, and manage everything in Formidable Forms via MCP — first-time setup and connecting a site, forms, fields, repeaters, entries, views, styles, form actions (email/confirmation/webhook), applications, PDF downloads, templates, and troubleshooting known MCP bugs

summerriversound Bundle

Use when writing an analysis, report, findings, or status doc for humans to read and share — not a spec an agent executes. Fights AI verbosity/slop and makes deep content easy to read.

technicalflight Bundle

把一张图片逆推成生成提示词（image to prompt）。看图后法证式重建它最可能的原始生成提示词，产出中文与英文两套重建提示词、4×2 风格标签、结构化 JSON prompt、recreation_prompt / prompt_core / negative_prompt，并对双语输出做语言桶校验。当用户说"反推提示词""这张图用了什么 prompt""图片转提示词""解析出这张图的 prompt""照这张图的风格帮我写 prompt""怎么做出这种图"时使用。

thecolonyai Bundle

Interact with The Colony (thecolony.ai) — an AI agent forum, marketplace, and DM network. Use for registration, posting, commenting, searching, marketplace tasks, service offers, marketplace reviews, Lightning tips, polls, webhooks, facilitation, 1:1 DMs and multi-party group conversations (with reactions, edits, pins, attachments), a personalised for-you feed, agent suggested-actions, notifications, achievements, content reports, agent-claim operator pairing, and profile management. Triggers on "colony", "thecolony", "post to the colony", "check the colony", "colony feed", "colony marketplace", "tip on colony", "review on colony", "colony group", "group DM".

jeph Bundle

Safely operate the `venmo` CLI for authorization status, balances, users, payments, requests, transfers, friends, activity, likes, comments, and reactions. Use when a user asks an AI agent to inspect or change their Venmo account, perform a Venmo action, or explain a venmo-cli command.

jeremylongshore Bundle

Analyze code changes and detect intent drift with the pr-to-spec CLI, converting a branch, staged edits, recent commits, or a GitHub pull request into an agent-consumable spec. Use when declaring intent before a change, checking completed work for drift, or producing review evidence. Trigger with "/pr-to-spec", "scan this diff", "check intent drift", or "declare change intent".

jiabaobei Bundle

当 Agent 接到专业任务（编码/爬虫/文件操作/API调用/数据分析/文档/部署/推送等）时，强制先查记忆层和技能索引，有匹配必用、无匹配必搜、答复时自动推荐（排除已装）。用于防止 Agent 跳过技能直接硬扛通用能力。跨平台通用（WorkBuddy/Claude/ChatGPT/Cursor/Gemini 等 20+ 框架）。完整版本史见 CHANGELOG.md。

jiahao-shao1 Bundle

远程 GPU 集群操作。当用户提到集群、远程执行、GPU、训练、同步代码时使用。触发词包括但不限于："连集群"、"同步代码"、"GPU 占用"、"集群上跑"、"remote bash"、"在服务器上"、"跑训练"、"看日志"、"tail log"、"mutagen"。即使用户没有明确提到集群，只要任务涉及远程执行或训练相关操作，也应该触发此 skill。

justcyl Skill

指导 agent 正确使用 pi-tmux 扩展管理后台 tmux 进程。涵盖 run/peek/send/kill/list 五个 action 的使用时机与参数选择、silenceTimeout 与 watchInterval 的配置策略、completion/silence/watchdog 三种自动通知的响应方式，以及常见工作流模式（dev server、build、交互式程序、长时间训练）。适用于 agent 需要在后台运行命令、监控进程输出、与交互式程序通信时。

jzocb Skill

Manage AI agent memory with P0/P1/P2 priority system and auto-archival. Use when: setting up memory management, cleaning up MEMORY.md, creating lessons files, configuring auto-archive cron, or reviewing memory structure. Reduces token usage by 70-80% while maintaining recall via semantic search.

kentaroajisaka Bundle

マネーフォワードクラウド会計の公式MCP（mf-official）を使用する際のガイド。認証フロー、仕訳取得・登録、試算表・推移表の取得、科目マスタの扱い、金額計算の注意点など、MF公式MCPの全ツールの使い方とクセを網羅。「MFの仕訳」「マネーフォワードの試算表」「MF認証」「MFで仕訳登録」等のキーワードで発動。

kings0527 Bundle

Evidence-first structural research for industries, companies, commodities, tradable assets, and policy systems. Use when the user requests StructFlow or needs a rigorous, source-grounded analysis of system boundaries, variables, causal drivers, feedback loops, nonlinear dynamics, regimes, market distortion, structural signals, or asset mapping. Run canonical input resolution, staged L0-L7 modeling, adversarial challenge, contradiction search, deterministic research gates, and auditable report generation. Use the host agent's model for reasoning; no separate LLM API key is required. Optional Tavily and AnySearch providers can supplement the host agent's own search tools.

klarkxy Bundle

Use zhihu-search for Chinese-community research: Zhihu links, real user experiences, product reputation or pitfalls, Chinese user opinions, domestic hot topics, and Chinese sources needing verification. Trigger for “知乎/知乎链接”“真实体验/口碑/避坑/大家怎么看”“国内用户观点/中文社区”“国内热点/最近在讨论什么”“查中文来源/核实中文信息”, even without naming Zhihu, plus setup or troubleshooting. Run one narrow on-demand CLI query; reuse matching Zhihu MCP tools only when visible. Do not use for repository-local code questions, pure math or logic, translation, or transformations of user-provided content unless Chinese-community evidence is explicitly needed. User data, knowledge bases, PDF/PPT, and OAuth require an explicit request.

kolibril13 Bundle

Build Blender node trees (geometry nodes, shader nodes, compositor) programmatically with the nodebpy Python library, executed via the Blender MCP. Use when the user wants to create or modify Blender node setups, geometry nodes, shaders, or compositor trees. Do not use for generic scene inspection or scene summaries; call the configured scene-summary MCP tool directly.

kwf18423059038-pixel Bundle

Orchestrate end-to-end Chinese law-firm bids in compatible agent environments. Use for 招标、采购、询价、磋商及投标/响应文件: guide beginners or professionals through tender-first intake; extract mandatory, scoring, pricing and submission requirements; assess private personnel and performance evidence; draft non-litigation, recurring-counsel, compliance, litigation or arbitration plans; inherit approved templates; create clean and review copies; assemble review indexes, signing and sealing materials; verify returned execution files; and deposit approved private practice lessons. Do not use as a source of current facts, laws, cases, qualifications, prices, signatures or seals.

labyla Bundle

Use when an agent needs to perform Twitter/X research with the local xsearch CLI, including checking auth status, importing user-provided cookies, searching posts/users, retrieving posts by ID, and handling JSON-only command output without exposing cookie secrets.

lalal-zzz Skill

One-stop toolkit to audit, optimize, and publish MCP servers and Agent Skills. Use when the user asks to inspect, audit, score, optimize, or improve an MCP server / Agent Skill project, or to package and publish it to NPM, Smithery, Official MCP Registry, GitHub, cursor.directory, Dify Marketplace, OpenAI GPT Store, Coze Store, FlowGPT, PromptBase, and popular awesome-lists.

leeguooooo Bundle

macOS WeChat CLI + local HTTP bridge + Wechaty Puppet gRPC gateway — send messages, query sessions / contacts / chat history / images / favorites, and expose stable HTTP / gRPC surfaces for agent integration. Use when the user asks to 'send a WeChat message', '发微信', query WeChat contacts/groups/messages, look up who said what in a chat, fetch images from history, export chat history, wire WeChat into Hermes / n8n / Dify / LangChain, or run any wechaty bot on a real macOS WeChat account. Uses the installer-managed, isolated WeChat 4.1.9 clone on macOS (Apple Silicon) and a `wechatuse_` activation code. One-time `wechat-use init` extracts the DB key; no sudo, no re-signing WeChat.app. Optional remote bridge — `wechat-use tunnel setup --hostname YOUR_HOSTNAME` exposes the local REST API via Cloudflare Tunnel for remote services to call.

liqqi77 Bundle

Govern versioned product delivery around one user-visible mainline, proportionate roles and gates, independent quality evidence, human acceptance, and bounded recovery. Use when organizing multi-stage development, task boards, agent collaboration, reviews, rework, releases, or a drifting or over-governed project. Do not use to add ceremony to an ordinary isolated edit that has no existing project gate.

liuyunlin Bundle

引导用户在 AI Studio 无代码训练平台完成大模型微调，支持 ERNIE 和开源模型（Qwen/LLaMA）。当用户提到微调、训练模型、SFT、fine-tune、问答对、让模型学会、定制领域助手、专属领域 AI、行业术语、调教模型、业务知识注入、领域适配、医疗问答、客服问答、无代码训练、AiStudio 训练、ERNIE 微调、模型训练、finetune 时，优先触发。

louisbrulenaudet Skill

Update locked agent skills from skills-lock.json one at a time. Use when the user asks to update, refresh, or sync project skills via npx skills update.

lout33 Skill

Use when working with agentmux, tmux-backed agent supervision, project managers, project workers, the local agentmux web UI, or durable terminal agent runtimes. Trigger on requests like create a worker, start a manager, supervise agents, attach to a worker, send input to an agent, import a tmux session, open the dashboard, or coordinate terminal agents in one repo.

lovstudio Bundle

Move a project folder AND migrate all its Claude Code state in one shot — session store, prompt-up-arrow history, running-session records. Supports both directory-level moves AND session-level cherry-picking (by id, regex on first user prompt, or interactive picker). Use whenever the user wants to rename/move a project directory and keep `claude --resume` working, or wants to move a subset of chats from one project to another. 移动/重命名项目目录并迁移所有 CC 历史，或按 session 粒度把某个话题的对话搬到另一个项目。

lovstudio Bundle

按用途生成图像、制作可编辑图文布局或整理图像提示词。支持明确输入与结果回读。Use to create an image, designed graphic, or image prompt.

derek-zhuolin Bundle

Turn a local talking-head / 口播 video into metadata, transcript, and an AI-composed card-based video where the agent composes cards by cloning styled templates and filling them (designing bespoke only when content needs it), then renders to MP4. Ships with the Interflow AI club sign-off ("Interflow AI 出品"). Use when the user asks for interflow-video-cut, 口播成片, talking-head repurposing, video takeaways, transcript cleanup, or turning a spoken video into a captioned card video.

derisk-ai Bundle

AI驱动的微服务故障根因分析技能，通过系统化的遥测数据分析流程，精准定位故障根本原因

diamitani Bundle

This skill uses two MCP tools from the `session_info` server to access today's Cowork sessions. **Purpose**: Returns a list of all Claude Cowork sessions. Use this skill when working with mcp guide tasks or workflows.

doniyor117 Bundle

Write prose that reads like a person wrote it, not a language model. Use this skill for any writing or rewriting task where a human will read the result as finished work or as the user's own words: emails, CVs and cover letters, applications, essays, articles, blog and social posts, READMEs, documentation, reports, proposals, product copy, messages. Also use it when asked to fix, tighten, de-slop, or humanize existing text, when the user says something sounds like AI, corporate, generic, or fluffy, or when they ask for writing that sounds like them. Trigger even when the user names no style, since nearly every writing request benefits from this.

duhu2000 Bundle

供应商分级、评分与评估 SKILL · 企查查 MCP V2.0 增强版。 对供应商进行 Kraljic 矩阵分级、6 维度评分与综合评估，适用于供应商准入审批、年度复核、退出决策等场景。 触发关键词：供应商评估 / 供应商分级 / Kraljic 矩阵 / 供应商等级 / 瓶颈供应商 / 战略供应商 / 供应商复核 / 供应商准入 / 新供应商审批 / 供应商审计 / 年度复核 / 供应商记分卡 / 供应商资质 / 供应商尽调 / 供应商健康度 / 供应商绩效复核 / 供应商退出 / 供应商风险画像。 适用场景：当任务是要把供应商**分级**到（战略 / 策略 / 商品 / 瓶颈）某一类、对其做**多维度评分**、或对供应商**准入 / 复核 / 退出**做评估时使用本 SKILL。 不适用场景：持续风险信号监控（请用 supplier-risk-qcc）/ 发票核对（请用 invoice-reconciliation）/ 物流绩效复核（请用 logistics-brief）/ 品类支出分析（请用 spend-analysis）。