AI & ML

sakhadib Bundle

Draft, expand, revise, structure, and maintain a full Islamic University of Technology BSc thesis from source documents using the local iutbscthesis LaTeX template. Use when writing or editing thesis metadata, front matter, abstract, chapters, related works, methodology, results, discussion, conclusion, citations, appendices, figures, tables, code listings, or compile guidance for main.tex. Also use when the user expects an AI agent to turn papers, notes, PDFs, LaTeX sources, bib files, data tables, and assets into a substantial thesis manuscript chapter by chapter without requiring repeated babysitting.

ssshooter Skill

Implement real-time streaming mindmap rendering using Mind Elixir in web applications. Supports streaming text parsing and incremental updates.

latitude-dev Bundle

Add or review Latitude Telemetry for LLM apps and agent harnesses. Use for Latitude tracing, LLM observability, missing traces, OpenTelemetry/OTLP integration in TypeScript, Python, and other runtimes, any documented provider or framework (OpenAI, Anthropic, Bedrock, Vertex, LangChain, LlamaIndex, Vercel AI SDK, Pydantic AI, Mastra, and more), and harness plugins for Claude Code, Hermes, OpenClaw, Pi, and Prime Intellect. Covers instrumenting a new project under an existing Latitude account (invoked directly) as well as the latitude-setup hand-off, discovering config via the Latitude MCP, and verifying that real traces land via the Latitude MCP, CLI, or API.

microsoft Skill

Alias for the /promptkit skill. Use this when the user wants to assemble a task-specific prompt using PromptKit, or says "bootstrap" to start the PromptKit composition engine.

microsoft Skill

Create custom agent capabilities when discovering novel tools, receiving task-agnostic tips from reviewers, or after researching specialized workflows not covered in existing instructions. Teaches structure, YAML optimization for LLM discoverability, and token efficiency.

paleo Skill

Conventions for writing sharp, concise prose and Markdown, for human readers and AI agents. Read before writing or editing documentation, READMEs, instructions, agent skills, or any prose text.

postplusai Skill

Decode benchmark videos, contact sheets, frames, or rough ideas into reusable opening structure. Use when a reference exists and the goal is to extract hook essence, viewer question, visual grammar, and drift boundaries before prompt writing.

srid Skill

Drive one AI agent from another through kolu's terminals: spawn a Claude Code / Codex / opencode session in a PTY, prompt it, watch the screen for its reply, read it, and prompt again — a create→send→snapshot loop run with the `kaval-tui` CLI directly (no MCP). `kaval-tui` writes input, reads scrollback, and provides a hook-free done-signal (`wait --until idle:<ms>` on raw PTY output); `pulam-tui` adds a precise agent-state done-signal when you drive hooked terminals. Triggers on "drive another agent", "send a prompt to a terminal agent", "have one agent prompt another", "agent drives agent", "orchestrate agents in terminals", "make Claude drive Codex", "prompt the agent running in that terminal", or wiring a loop where one coding agent supervises another.

ojpalenzuela Bundle

Generate a complete, project-specific AGENTS.md compatible with the agents.md standard (60k+ open-source projects). Also generates platform-specific files (.cursorrules, Copilot instructions, GEMINI.md) for detected platforms. Use when setting up AI agent rules, creating project conventions for Claude Code, Cursor, Copilot, OpenCode, or any agent that reads AGENTS.md. Supports three modes: full (rich AGENTS.md + companion rule files + multi-platform output), minimal (single 30-line AGENTS.md following the agents.md standard), and update (diff existing, regenerate only changed). Detects monorepos for nested files. Extracts design tokens from config files. Includes confidence scoring, managed blocks, audit logging, and state memory. Every command and convention is derived from the project's real package.json, config files, and directory structure — never generic templates.

brycewang-stanford Bundle

Recommend AND run open-source AI tools, agents, Claude Code / Codex skills, and MCP servers for any stage of a literature review — searching, reading, extracting, synthesizing, screening, citation-checking, and paper writing. Use when the user asks "what tool should I use to..." OR "install/run/use <tool> to ..." for research/lit-review work: automating a survey or related-work section, PDF→Markdown extraction for LLMs (MinerU/marker/docling), PRISMA / systematic review (ASReview), citation-backed Q&A over PDFs (PaperQA2), wiring papers into Claude/Cursor via MCP (arxiv/paper-search/zotero servers), or chatting with a Zotero library. Also answers direct literature lookups — find papers on a topic, resolve a DOI, get an open-access PDF, check a citation, search PubMed / OpenAlex / Crossref / Semantic Scholar / Europe PMC / arXiv — with no install and no API key. Ships a launcher (scripts/litrun.py) that installs each tool in an isolated venv and runs it, plus keyless bundled scripts for multi-source search and

b18797245781-commits Bundle

1688选品库MCP服务引导技能。当用户需要通过1688平台进行商品选品、爆品筛选、供应商评估、商品销量分析、热销榜单查询、类目浏览时触发此技能。本技能引导AI正确理解用户选品意图、选择合适的MCP工具、补全关键参数、并以视觉化方式（图片+表格+卡片）展示选品结果。触发本技能后，所有1688选品相关工具调用默认使用店雷达（dld MCP）连接器，工具包括 mcp__dld MCP__product_search_list、mcp__dld MCP__product_billboard_list、mcp__dld MCP__product_info、mcp__dld MCP__get_category_info、mcp__dld MCP__get_user_info。

bablobanov Bundle

Create/read/append/toggle/rebuild a native Telegram To-Do list (a real checklist with tappable checkboxes and a progress counter) in allowlisted chats or forum topics via a Telethon user session. Use when the user asks for a Telegram task list, a checklist with checkboxes, a To-Do list, to collect tasks or ideas from a chat/topic into a checklist, or to tick/add/rebuild list items. Supports a research contract: the agent collects candidate tasks from real chat messages into plan.json and the script validates it (allowlisted target, source links present in task texts, duplicates rejected, limits) before anything is sent. This is NOT markdown and NOT an internal todo - it is a real interactive Telegram object. The Bot API cannot post checklists into groups or topics (business-account private chats only), hence a Telethon user session. This is a WRITE skill: act only on an explicit user request, only into allowlisted chats, and always plan/dry-run before the real write.

baranskyi Bundle

Adversarial truth-extraction agent that interrogates the user to get past their own self-deception, evasions, and comfortable cover stories to what is ACTUALLY true. A harder-edged sibling of thinking-partner: instead of gently facilitating, it treats the user as a subject with a story to crack open — using the real elicitation craft of intelligence and police interrogation (rapport, the Scharff method, strategic use of evidence, contradiction-pinning). Use when the user says: 'допроси меня', 'interrogate me', 'колись', 'вытащи из меня правду', 'допрос', 'расколи меня', 'press me', 'grill me', 'проверь меня на прочность', 'я себе вру', 'help me stop bullshitting myself', or wants to stress-test a decision, claim, or plan by being questioned hard. Also trigger when the user is obviously rationalizing, hiding the real reason, or asking to be held to account. NEVER hands over conclusions — it extracts them from the user under questioning. Refuses to coach interrogating anyone but the consenting user.

benjaminstelzer Bundle

Ask Claude Code and a fresh normal Codex SOL project task in parallel for independent read-only opinions, reviews, critiques, comparisons, or alternative analysis. Use when the user explicitly asks for both Claude or Fable and SOL, says "Frage Fable und SOL", or invokes $ask-claude-and-sol-for-codex. Do not use this Skill instead of ask-claude-for-codex, ask-astra-for-review-for-codex, or ask-claude-and-astra-for-codex, or for a generic request for another model or two opinions.

beriki770-ship-it Bundle

Drive Blender directly through the connected Blender MCP server (mcp__Blender__* tools) — inspect scenes, edit objects and materials, run bpy code, render, screenshot the viewport, and analyze .blend files. Use this skill whenever the task touches Blender itself - the user mentions Blender, a .blend file, modeling, materials, lighting, rendering, a 3D scene to build or fix, or asks "what's in this blend file". Also use it when a web-3D task needs a source model prepared or rendered inside Blender first, before any glTF export step.

blankyname Bundle

Use the local ChatGPT Web prompt-forwarding MCP bridge: submit prompts and reference images to an already-open real Chrome ChatGPT tab via CDP.

blavkgokuvnn Bundle

Track durable phase state for long, multi-phase work with a minimal StateM runbook. Use only when the active surface has a local Python environment and writable project filesystem, and the work has at least three substantial phases or durable mutable execution state that may outlive the current context. Do not invoke on web or mobile surfaces, in environments without local Python/filesystem access, or for short tasks and ordinary checklists.

blongworth Bundle

Query and analyze New England Shelf Long-term Ecological Research (NES-LTER) oceanographic data with the bundled command-line client or the optional NES-LTER MCP. Use for NES-LTER stations, cruises, casts, CTD profiles, underway or bottle data, nutrients, HPLC, chlorophyll, events, cruise tracks, schemas, variable resolution, and source-backed summaries.

buildaloud Bundle

Skills Marketplace MCP Broker

camolechowski Bundle

Read, search, send, poll, and wait on iMessages on macOS via the imsg CLI. Use when asked to check texts/iMessages, message someone, wait for a reply, send a file over iMessage, or monitor a conversation from a script or agent. Requires macOS with Full Disk Access granted to the terminal.

campfirein Bundle

Persistent, portable memory for AI builders and agent teams. Use BEFORE any non-trivial work to retrieve prior decisions, patterns, and gotchas; use AFTER finishing to record what was learned. ALSO use to authenticate with ByteRover — when the user says to authenticate / auth / log in / sign in / connect / link ByteRover (or log out / sign out / disconnect), run the auth flow. Iron Law: query before you think, curate after you implement.

capsrock Bundle

Xcode 27 の iOS シミュレータをコマンドと MCP で操作・検証するときの道具の選び方と落とし穴。Device Hub を開かずに済む経路と、開くしかない操作の区別。タップ・スワイプ・撮影・回転・要素ツリーの取得、状態の作り込み、xcodebuildmcp / axe / simctl / devicectl / Device Hub / Xcode MCP の使い分けを扱う。シミュレータで画面を確認したい、タップが効かない、要素の座標が取れない、回転させたい、スクリーンショットを撮りたい、UI 自動操作が不安定、Device Hub を開かずに済ませたい、といったときに使う。

carbonteq Bundle

Operate ShareView review workspaces with the bundled CLI. Use when an agent needs to discover or create projects, upload or download artifact versions, inspect or create version-scoped comments, respond to review feedback, or decide between the ShareView CLI and MCP in a network-restricted environment.

chengyansen-ai Bundle

Plan, audit, configure, run when authorized, evaluate, and package MiniMax H3 LoRA training for images, video, stereo audio, motion, and true Ref2VA reference conditioning. Use when a user wants an auditable H3 LoRA dataset or training run; do not use for prompt-only Hailuo generation or unrelated image-model LoRA training.

choice-form Skill

READ THIS BEFORE calling any Lovision MCP tool. Use whenever the Lovision MCP server is connected or the user mentions Lovision, lovision.ai, mcp.lovision.ai, Lovision Desktop Local MCP, a Lovision project/document/canvas, Blueprint design generation, project operations, quality checks, preview/export, or asks an external agent to create, inspect, modify, or hand off native Lovision design work. Always initialize with lovision.session.init first, resolve the target before writing, read document context and capabilities before edits, prefer Blueprint for structured design work, run quality and preview after meaningful writes, and finish with lovision.session.finishWork.

chrischall Bundle

The example skill shipped with skill-mcp — shows the SKILL.md shape, a bundled reference file, and a declared runnable script.

chrisliu298 Bundle

Dispatch multiple independent agents to answer the SAME complete question from different analytical lenses, then synthesize. Reach for this by default, without waiting to be asked, when redundant cross-model judgment could change the decision: ambiguous architecture/design tradeoffs, high-stakes or hard-to-reverse changes, competing root-cause hypotheses, or failure-mode-sensitive reviews — and whenever you would otherwise spawn 2+ independent reviewers for one question. Skip for trivial lookups, deterministic transforms, routine small edits, mechanical syncs, and single-correct-answer tasks. With no leading number, autonomously decide N and gpt-pro from the question every time (N=1 is the bottom-rung anchor, never a lazy default); with an explicit number (e.g. `prism 2 1`), honor it verbatim and skip auto-sizing. There is no reasoning-effort knob. Scale above the anchor only when decision risk justifies the 7N-agent cost.

chrisliu298 Bundle

Guide for writing and refining prompts for Claude or GPT/Codex models, distilled from each vendor's official best practices. Use this skill whenever the user wants to write a new prompt, improve or debug an existing prompt, or learn model-specific prompting techniques. Also trigger when the user mentions "prompt engineering", "system prompt", "Claude prompt", "GPT prompt", "Codex prompt", "OpenAI prompt", or asks how to get better results from any LLM. Accepts an argument: "claude" for Claude/Anthropic guidance, "gpt" for GPT-5.5 guidance, "codex" for Codex agent guidance.

chrisliu298 Bundle

The ONLY way to call Codex, DeepSeek, MiMo, GLM, or Grok. Use whenever the user wants to ask, delegate to, or get a second opinion from Codex, DeepSeek, MiMo, GLM, or Grok. Do NOT run the codex, deepseek (ds), mimo (mm), glm, or grok CLI directly — from the main agent or a subagent; always use this skill's relay call command. Triggers on "ask/have/send to/get/delegate to codex" or the same with "deepseek"/"mimo"/"glm"/"grok", "second opinion", "relay".

chujianyun Bundle

Prompt engineering expert that helps users craft optimized prompts using 57 proven frameworks. Use when users want to optimize prompts, improve AI instructions, create better prompts for specific tasks, or need help selecting the best prompt framework for their use case.

civilization-os Skill

Manage custom Skill directories and Model Context Protocol (MCP) server configurations (stdio, streamable-http, sse) in DeepSeek Harness.

clasen Skill

Instructions for using the ModelMix Node.js library to interact with multiple AI LLM providers through a unified interface. Use when writing code that calls AI models (OpenAI, Anthropic, Google, Groq, Perplexity, Grok, Moonshot, MiniMax, Fireworks, Together, Lambda, Cerebras, OpenRouter, Ollama, LM Studio), chaining models with fallback, getting structured JSON from LLMs, adding MCP tools, streaming responses, managing multi-provider AI workflows, round-robin load balancing, or rate limiting API requests in Node.js. Also use when the user mentions "modelmix", "ModelMix", asks to "call an LLM", "query a model", "add AI to my app", or wants to integrate any supported provider.

claude-skills-kit Skill

Consult a cold-context advisor subagent — Fable 5 or a fresh Opus — for a plan, a diagnosis, a critique, or a decision, while keeping Fable quota usage low. Use when the user says "ask fable", "ถาม fable", "ask opus", "ถาม opus", "consult an advisor", "ปรึกษา advisor", "agent-advisor", or asks for a second opinion / architecture plan / debugging strategy / critique of an existing plan on a genuinely hard, high-stakes, or hard-to-reverse problem. Runs a gate first to decide whether the task needs an advisor at all (an explicit user request always passes the gate), picks the tier and the consult mode, hands it a curated brief (not the raw transcript), checks the answer against the user's explicit instructions, and routes implementation back to the current main model. Do NOT invoke for small, clearly-scoped, low-risk work — the gate exists precisely to keep those off the advisors.

cleudersons Skill

Use when the user wants Claude to publish posts, edit pages, or manage a self-hosted WordPress site directly — connecting Claude Code (or any MCP client) to a WordPress blog via an MCP server, especially when "connect Claude to my WordPress/blog", "let Claude publish articles", or "manage my site with AI" comes up.

cogine-ai Skill

Guide for using the clawpacker CLI to export, import, inspect, and validate portable OpenClaw agent packages. Use this skill whenever the user wants to migrate, share, clone, package, export, import, restore, or move an OpenClaw agent between instances. Also use when the user mentions clawpacker, .ocpkg, agent portability, agent templates, workspace packaging, or asks how to transfer an OpenClaw agent to another machine or environment. Even if the user just says "package my agent" or "move this agent" or "set up a new instance from an existing agent" — this skill applies.

cogpros Bundle

Two-level system audit. Level 1: inventory everything, score health, output manifest. Level 2: design a custom event bus topology from the manifest. For setting up new agent systems or wiring existing ones into a bus.