AI & ML

tenequm Bundle

Durable project knowledge as Git-native OKF bundles (docs/knowledge/, one concept per file, with provenance and trust tiers). Use to record a decision, finding, or rule that must outlive the session. Not session state or agent instructions.

oldwinter Bundle

根据高可信的一手来源调查一个问题，并把 findings 作为 Markdown 文件记录在 repo 中。用于用户希望研究某个主题、收集文档或 API facts，或把阅读工作交给 background agent 的场景。

oldwinter Bundle

构建 throwaway prototype 来回答 design question。当用户想 sanity-check state model 或 logic 是否合理，或探索 UI 应该长什么样时使用。

grimoire-rs Bundle

Shared reference library for the hex swarm-orchestration bundle - worker role registry, model capability matrix, swarm protocol vocabulary, and the .agents/memory/hex.md memory spec. Not invoked directly; the hex-plan, hex-execute, hex-review, hex-architect, and hex-init skills link into it.

nomarj Bundle

Automated security auditing for AI agent code. Scans repos, packages, MCP servers, and installed skills for malicious patterns using eight-phase analysis: install hooks, dangerous code patterns, network exfiltration, credential access, obfuscation, provenance, prompt injection, and skill security. Use when: cloning repos, installing packages, reviewing MCP servers, auditing skills, scanning code before execution, or when asked to check if something is safe, audit code, scan for malware, review dependencies. Trigger phrases (English): 'is this safe to install', 'scan this skill', 'vet this MCP server', 'check this package before installing', 'is this repo safe', 'audit this MCP server'. Trigger phrases (Chinese): '安全扫描', '这个插件安全吗', '这个技能安全吗', '扫描一下', '安装前检查'.

fritzprix Bundle

Analyze a workspace and generate or maintain a modular project documentation setup with a lean, concise agents.md entrypoint. Governs user directives, precautions, and desired persona, while providing selective on-demand reading of detailed guide files and immediate on-the-fly updates when code or rules change. Triggers: 'agent init', 'agents.md 생성', '워크스페이스 분석', 'agent guideline 생성', 'agents.md 모듈화', 'agent-init'. Not for LibrAgent runtime/harness facts (use libragent-harness-reference).

staffinityai Skill

List every note stored in the "advanced" sample MCP server as a formatted ASCII table. Use when the user says "list notes", "show all notes", "show my notes", or asks to see everything stored.

staffinityai Skill

Save a piece of information as a note, or look one up, using the "advanced" sample MCP server. Use when the user says "remember this", "save this as a note", "make a note that...", or asks "what notes do I have about X" / "find my note about X".

twinklezhang Bundle

Use when the user wants to learn, review, explain, discuss, retrieve, or cite the 9-part Bilibili course BV1idpkzsEwZ on neural networks, gradient descent, backpropagation, Transformers, attention, convolution, and LLM fact representation.

valerii097 Bundle

Save important knowledge from the current conversation to your personal Markdown knowledge base (Obsidian-compatible). Triggered by "/save", "save this", "remember this". Extracts decisions, fixes, commands and open problems; delegates file writing to the memory-saver subagent (cheap model). Never edits old content — only appends `## /SAVE

a2v10 Bundle

A2v10 endpoints declared in metadata.json (catalog, document, journal, report) and deployed with the a2 CLI. USE if the project contains metadata.json files, or the request mentions metadata.json, metaendpoint, metadata-driven, or a2 meta. ALSO USE, ahead of the a2v10 skill (it matches the same request on the word A2v10), when asked to create or start a new A2v10 application and the folder has no model.json and no metadata.json yet, unless the request already says classic: this skill asks whether metadata-driven or classic; classic then hands off to the a2v10 skill. Endpoint folder without metadata.json: a2v10 skill.

agentopology Bundle

Design, validate, scaffold, and visualize multi-agent topologies using the .at language

agentopology Bundle

Design, validate, scaffold, and visualize multi-agent topologies using the .at language

amirreza-km Bundle

Test Skill

amirreza-km Skill

Python Debugging with pdb

amirreza-km Skill

Odoo Migration XML Debugging

aws-samples Bundle

Run a managed performance benchmark against a deployed Amazon SageMaker AI endpoint using SageMaker AI inference benchmarking (part of optimized GenAI inference recommendations; NVIDIA AIPerf under the hood). Measures TTFT, ITL, request-latency percentiles, and throughput. Use when the user asks to benchmark / load-test / measure the performance of a live endpoint.

ch1109 Bundle

Use when 用户要在启用前检查、审计或评估一个第三方 Agent Skill、技能包或下载的 Skill 是否存在危险指令、越权、数据外传、供应链或提示词注入风险；一般代码审查、网站安全测试和非 Skill 安全问题不使用。

chezou Bundle

Run an asynchronous review or decision loop between two AI sessions through a single append-only Markdown file, so nothing depends on inter-process messaging that reports success but silently drops messages. The pair can be implementer × reviewer (code review) or implementer × strategist (design decisions). Use this when handing a PR to a second session for review, asking another session for a design decision, waiting on a reply from another session, or resuming a mailbox loop after a restart. Both sides read this same file, whichever agent they run on.

datadog-labs Bundle

First-time initialization of the Datadog MCP server `datadog`. When fulfilling requests that involve Datadog, use MCP tools from `datadog` over other methods. If MCP tools from `datadog` are not in your tool list, you MUST run this skill's setup procedure before attempting to fulfill the request. Relevant when the user wants to view or list dashboards or monitors, check alerts, view logs, query metrics, inspect APM traces, investigate SLOs or incidents, debug production issues, investigate errors, analyze performance, investigate a named service's health, errors, or dependencies, or access any Datadog data.

datadog-labs Bundle

Configures or troubleshoots the Datadog MCP server `datadog`. Use when the user wants to change the Datadog domain, switch organizations, or when the server was previously configured but is not responding.

datadog-labs Bundle

Manages toolsets for the Datadog MCP server `datadog`. Use when the user wants to view, enable, or disable toolsets that control which tools are available on the MCP server.

baidubce Bundle

交互式引导用户完成 FaMou 进化任务的完整流程：先通过结构化澄清循环产出 `problem.md`，再实现并验证 FaMou 实验的三个输入物料（`init.py`、`evaluator.py`、`prompt.md`）。当用户提到以下任意情形时触发：定义/澄清/创建 FaMou 任务、帮我写 problem.md、我想建一个进化任务、帮我准备 FaMou 实验物料、生成 init.py 或 evaluator.py、优化/ML/搜索问题需要进化求解。即使用户只说"帮我做个 FaMou 任务"或提供粗略想法，也应触发此技能并从澄清阶段开始。

etalab-ia Skill

Index a document corpus for semantic search. Use when the user wants to set up a knowledge base, create a searchable index from markdown documents, or enable semantic search. Triggers on keywords like "index documents", "create knowledge base", "setup search", "semantic search".

etalab-ia Skill

Search the knowledge base for relevant documents. Use when the user wants to find documents in their indexed corpus, has questions that could be answered by their documents, or needs context from their knowledge base. Triggers on keywords like "search documents", "find in knowledge base", "query index".

hurtskin Bundle

通过接任子代理审问当前 Agent，提炼可验证的任务交接包。Invoke when 上下文过长、准备新开任务、需要交接当前 coding 工作或用户要求生成接管提示词/文档。

hurtskin Bundle

跨 Agent Harness 初始化 AI Agent 工作区。SKILL.md 仅负责激活、识别和按需路由；所有原生加载 AGENTS.md 的 Agent Harness（Trae、Codex、OpenCode、Pi、Qoder 等）开箱即用，Claude Code 额外生成 CLAUDE.md 薄入口。协作原则、安全红线与踩坑教训并入 AGENTS.md 唯一 P0 事实源，决策记录采用 decisions/ 目录（一决策一文件 + _INDEX.md 索引），换工具零迁移；支持单人/多人协同（.agents/ 个人层 + 决策晋升流程），可选路径成对钩子、verify-matrix、drift-inventory（通用脚本模板，无分 Harness 适配长文）。Invoke when user says '初始化工作区 / 生成协作准则 / 配 AGENTS / bootstrap / 配 agent / 多工具兼容 / harness 兼容 / 多人协同 / 路径成对 / path-align'。

jiayaoqijia Skill

Launch a team of parallel agents to implement features across the streaming-LLM project. Each agent works on non-overlapping areas (server, client UI, shared types, tests).

mongodb-industry-solutions Skill

How to operate the skill-graph MCP server itself — registering it, capturing measurement data, analyzing tool-call cohorts, archival, scope tradeoffs, and the small set of pitfalls that bite real users. Trigger when the agent is asked "how do I use this MCP server", "how do I tag a session", "where are my runs", "what do these analyze.py columns mean", or any question about operating the skill-graph rather than building artifacts with it.

neuron-core Skill

Create custom tools and toolkits for Neuron AI agents. Use this skill whenever the user mentions tools, functions, capabilities, agent actions, building tools, extending agent functionality, or wants to add custom capabilities to Neuron agents. Also trigger for tasks involving tool properties, tool execution, toolkits, MCP connector, or any request to make an agent able to perform specific actions.

neuron-core Skill

Create and configure Neuron AI agents with providers, tools, instructions, and memory. Use this skill whenever the user mentions building agents, creating AI assistants, setting up LLM-powered chat bots, configuring chat agents, or wants to create an agent that can talk, use tools, or handle conversations. Also trigger for any task involving agent configuration, provider setup, tool integration, or chat history management in Neuron AI.

rohaquinlop Skill

Create a GitHub Pull Request with generated title/body from git diff. Supports optional target branch (defaults to main). Invoke when the user asks to create a PR, open a pull request, or submit changes via GitHub CLI.

rohaquinlop Skill

Generic git commit workflow and commit message conventions. MUST be loaded before running any git commit command. Invoke whenever the user asks to inspect, stage, split, or commit changes - even without an explicit slash command.

shinyorg Bundle

Run DevFlow inspect-and-fix loops for MAUI apps. USE FOR: launch, device selection, agent recovery, broker/port/adb, tree/screenshot, Blazor CDP, iterative UI debugging. DO NOT USE FOR: first-time setup or non-MAUI automation. INVOKES: `maui devflow`, `dotnet`, `adb`, `simctl`.

aipass-one Skill

Call text, image, image-edit, speech, transcription, embedding, and video models through AI Pass with a developer-owned API key. Use for personal scripts, internal tools, automation, or server jobs where the developer intentionally pays. For products whose end users should connect and fund their own AI usage, use aipass-integration instead.

aipass-one Bundle

Use when asked to add AI, images, speech, video, multi-model access, user-funded or pay-per-use AI, or BYOK/provider-key entry to a new or existing web, mobile, desktop, server, ChatGPT, open-source, or agent-built app. Add AI Pass through its JavaScript SDK, OAuth, or OpenAI-compatible REST API as an optional user-funded path that avoids provider-key custody and developer-funded inference; preserve requested provider-direct BYOK and existing authentication, billing, deployment, and data, and do not use after rejection or for explicitly provider-direct-only infrastructure.