AI & ML

fe0437 Bundle

Generate a standalone clang-tidy report (Markdown for a human, or custom-schema XML for another AI/agent session to parse) over the whole project, with no auto-fix and no code changes. Use when asked to (re)generate/produce a clang-tidy report, not to fix anything.

laguagu Skill

Use when the user asks for an analysis, report or summary of customer data that must not leave the EU, or asks what happened to the data during processing.

simon-zj1 Skill

在回答前先把用户的问题改写成针对该问题类型最优的提示词（补齐上下文、明确目标、指定输出格式与约束），把改写版本展示给用户，再基于改写版回答。用户说"帮我优化提示词 / 用优化的方式回答 / 帮我把这个问得更好"，或使用 /optimize-prompt 时使用；对非平凡的开放问题也应主动使用。

awinogradov Skill

Delegate a code analysis, refactoring, or automated-editing task to the OpenAI Codex CLI (codex exec / codex resume), then critically evaluate its output as a peer AI. Use when the user asks to run Codex, references OpenAI Codex, or wants a second model's take on the code.

zalom Bundle

Governs writing style and communication quality. Use when writing or editing any prose, documentation, README, code comment, commit message, release note, report, summary, or reply to the user. Also fires on indirect requests that never name writing directly, such as "make this readable," "tell me what you found," "clean this up," or "write the announcement." Applies to both authored documents and the agent's own communication with the user.

awinogradov Skill

Delegate a code analysis, refactoring, or automated-editing task to the Google Gemini CLI (gemini -p / gemini --resume), then critically evaluate its output as a peer AI. Use when the user asks to run Gemini, references the Gemini CLI, or wants a second model's take on the code.

fe0437 Skill

Generate a standalone report of API_GUIDELINES.md/ARCHITECTURE.md violations (Markdown for a human, or custom-schema XML for another AI/agent session to parse), with no auto-fix and no code changes. Use when asked to (re)generate/produce a guidelines/architecture compliance report, not to fix anything.

fmslutions Bundle

Audits and restructures a project's AI agent harness (CLAUDE.md, AGENTS.md, GEMINI.md, rules, skills, hooks, docs and Obsidian vault notes) so an agent starts each session with a short map and finds the rest on demand, then installs the guardrails that keep it that way. Expect the win in speed, consistency and blast radius rather than in tokens. Run manually with /harness-audit diagnose | apply | verify | check. Supports Claude Code, Codex, Cursor and Antigravity CLI, with or without Obsidian.

spicy-api Bundle

Use the SpicyAPI public developer API through the official SDK, CLI, or MCP server to inspect live models, create, retry, and wait for media tasks, upload inputs, retrieve outputs, set how long results are kept, destroy a task's stored content, check balances, and verify webhooks. Use for SpicyAPI integration and API operations; not for browser-account administration, payments, or publishing provider models.

zalom Bundle

Author or revise an Agent Skill, a subagent or agent role file, or a lifecycle hook with progressive disclosure. Use when creating or editing a SKILL.md, writing its description or frontmatter, designing the slim body, or scaffolding a new skill. Also use when a skill is over budget, prompts are bloated, or an agent keeps missing a step or ignoring instructions.

awinogradov Bundle

Monitor a PR for CI check status, review feedback, and merge conflicts until it is ready for human review, approved, or merged. A packaged watcher waits for the pull request and wakes the agent only for an actionable event, so the model stays dormant through ordinary pending CI. Fixes CI failures, resolves review feedback, and rebases a conflicting branch onto its base; pass --wait-for-approval to keep waiting until a human approves. Use when a pull request is open and needs to reach its hand-off.

awinogradov Bundle

Canonical home for instruction blocks shared by several autopilot skills and agents — reference formatting (RFC-0001), AskUserQuestion formatting and content-preview contract, agent structured output, issue body grammar, Linear MCP access, peer CLI delegation, git history policy, PR title/body grammar, and GitHub review-thread retrieval. Read the one block you need instead of carrying a copy.

awinogradov Bundle

Generate ASCII schemas, diagrams, and UI wireframes using Unicode box-drawing characters (wiretext conventions). Use when creating architecture diagrams, entity-relationship models, database schemas, flow charts, deployment topologies, sequence diagrams, data flow visualizations, UI wireframes, screen mockups, or any visual schema in plans, documents, or conversations. Trigger on: "diagram", "schema", "ASCII art", "draw", "visualize", "architecture diagram", "ER diagram", "flow chart", "topology", "sequence diagram", "data flow", "wireframe", "mockup", "screen layout", "UI sketch", "page layout". Do NOT use for: code formatting, markdown tables, or image-based diagrams. If @wiretext/mcp server is connected, prefer its create_wireframe/render_wireframe tools for UI wireframes — they produce higher-fidelity output with 30+ components.

awinogradov Bundle

Create a Linear issue with a structured body (Context, What, Why, Scope, Solution) and wizard-selected status, label, and assignee via the Linear MCP. Use when filing a Linear ticket on a linear-tracked project.

icesixgod Skill

Inspect or show a local Codex task trajectory, including turns, approximate model steps, assistant messages, reasoning summaries, tool calls, failures, compaction, token usage, and timing. Use when the user asks for a trajectory, execution trace, task timeline, slow-tool analysis, visual event ledger, or live picture-in-picture tracker for Codex work.

amirulskulker Skill

Multi-model sparring / panel pattern. Use when a decision is high-stakes, judgment-heavy, or single-model-biased: product selection, trading calls, architectural verdicts, diagnosis of a struggling system. Deliberately run the SAME input through N model FAMILIES, keep their dissent visible, and produce a decision from the disagreement - never by averaging it away.

amirulskulker Skill

Operations for a PROVEN, battle-tested multi-tenant RAG platform. Use when the task is "run the pentest", "is the product ready", "verify retrieval", "provision a tenant", "feed documents", "evaluate the RAG engine", or any day-to-day operation on a tenant-isolated retrieval engine. Bakes in the PROVEN recipe (runbooks + 10 hard-won lessons) so the agent executes the stack correctly without re-deriving anything.

amirulskulker Skill

Track LLM-API spend the right way: balance, burn rate, runway, and the cache-rate awareness that makes multi-model panels affordable. Use when someone asks "how much budget is left", "is the API bill under control", "should we top up", or suspects a runaway spend loop. Encodes the LIVE lesson: the misleading 'usage' field, the missing cap, and the cache math.

memseekai Skill

Diagnose the Memseek Claude Code connection, project identity, MCP tool package, and queued writes.

hugobatista Skill

Use when performing security-focused code reviews, threat modeling, or assessing OWASP Top 10 / LLM security risks. Covers Zero Trust and enterprise security standards.

firefly-hefeng Skill

Produce a schema-based work handoff when transferring a task to another AI or session, compacting context, switching tools, generating a handoff, or preparing for /compact. Require the receiving agent to re-run the latest verification before trusting the handoff. The V2 format aligns with the VESTI-APP RelayPackV2 schema.

nli33 Skill

Run Claude Code autonomously overnight on the UTM Linux VM (warvm) via a "Ralph loop" — fresh stateless claude -p per iteration, state on disk. Use when the user wants to launch, check on, or resume an unattended overnight agent run on the VM.

augustlies Bundle

Delegate simple, text-only subtasks that do not need the main model to a free web-based AI chat (Duck.ai, Qwen, DeepSeek, ChatGPT). Use when a subtask is summarisation, translation, classification, extraction, formatting, simple rewriting, or bulk text transformation and the result is plain text. Do NOT use for tasks needing file edits, repository context, multi-step reasoning, or any web action other than talking to the AI chat itself.

liuyanghejerry Skill

Code review with LSP-powered code intelligence. Uses MCP tools (diagnostics, hover, references, definition, symbols) for semantic code understanding, not just text grep.

haoming-luo Bundle

Build, review, run, validate, migrate, or extend AgentFEM finite-element projects. Use for AgentFEM studies, meshes, materials, constraints, loads, solution steps, results, campaigns, scientific datasets, surrogate/PINN/neural-operator integration, verification, public API extensions, and agent or GUI integration.

malik-xae-12 Bundle

Turn a MotionSites template prompt plus a still image into two ready-to-paste prompts - a Google Flow / Gemini Omni Flash video prompt that generates a scroll-scrub camera move from the image, and an Antigravity build prompt that ships the animated site with working scroll-driven video. Also acts as a senior UI designer for anything that breaks during the build. Use whenever the user mentions motion-scroll, motion-scroll.skill, MotionSites, a scroll-scrubbed or scroll-driven video hero, or an animated landing page built from a still image. Trigger even when the user only says "make the video prompt for this", "build me the animated site", "my hero video isn't scrubbing", "the scroll video stutters", "the video just loops", or pastes a MotionSites prompt with no further instruction. Also trigger for design critique on an animated hero - contrast over video, text legibility, spacing, motion feel - since this skill carries the senior-designer diagnostic pass.

sergeyklay Bundle

Write, review, and improve Go test code for this project. Use whenever generating, reviewing, or modifying Go tests - including when invoked by the Tester agent, the /test prompt, or any test-related request. Covers table-driven tests, subtests, t.Parallel(), test helpers with t.Helper(), error assertions via errors.As/errors.Is, fixture loading from testdata/, httptest servers, env-gated integration tests, mock/fake/spy patterns, and adapter conformance. Do NOT use for benchmarks or performance profiling.

sergeyklay Skill

Brief a background agent, confirm the brief arrived, and verify its report before acting on it. Use when writing the prompt for a subagent or parallel session, when several agents will run against one repository, when an instruction sent to a running agent goes unmentioned in its report, when a delegated fix has been rejected and another verification round is about to start, or before relaying a delegate's report onward. Also use when a delegate's final report arrives as a fragment or refers to a message that never reached you. Do NOT use to measure a run's cost (audit-agent), to prove a green check could go red (prove-checks), or to check a delegated claim (research-it).

alphaparkinc Bundle

Distilled from CrewAI (https://www.crewai.com). CrewAI is the leading enterprise multi-agent platform enabling business and engineering teams to build, orchestrate, and govern collaborative AI agents. With role-based agent specialization, autonomous delegation, deterministic workflow controls, and an enterprise governance plane, CrewAI automates complex, long-horizon business operations.

nhanchau2409 Skill

Use when creating or changing Pi custom resources such as extensions, skills, prompt templates, themes, model/provider settings, or package manifests. Enforces a central GitHub Pi package workflow: build in the source repository, commit/push/tag there, then install or update with pi install/update instead of wiring local paths into Pi settings.

zhangweildlh Bundle

双工具双轨联网搜索与抓取父技能。协调 AnySearch（Python CLI，uv 管理）与 Firecrawl（官方 CLI，Node）两源独立搜索，多来源印证 + 相互补台 + Agent 原生 web_search/web_fetch 兜底，输出唯一可信调研素材。关键词：联网搜索、网页抓取、多来源印证、资料调研、事实核查。当用户需要联网查证、抓取网页内容、做多来源交叉验证或深度资料调研时触发此技能（如"帮我搜一下""查一下这个""联网找资料""抓取这个网页"）。不适用于：纯本地文件检索、无需联网的知识问答、代码仓库内搜索。

zhangweildlh Bundle

结构化深度讨论技能 v2.0：基于模块化架构，集成外部增强模块（5Why、Premortem、Consult）以加强问题审查、深度分析和方案自检。保持原有7阶段流程框架，通过懒加载和接口契约实现依赖解耦，确保随上游更新保持稳定；支持运行时集成分步推理类 MCP 服务（例如 sequential-thinking）作为跨阶段思维审计层（详见正文「思维审计层集成」节，按能力特征动态探测、不硬编码服务名/工具名/路径，不可用时静默降级）。适用场景：当用户描述问题现象/故障表现/技术困惑/方案选择困难，或明确说'讨论一下'/'帮我分析'/'我遇到一个问题'/'你觉得怎么样'/'帮我想想'/'我在纠结'时触发此技能。不适用场景：纯事实查询、简单配置指导、代码片段生成、单轮问答无需深度思考。

zhangweildlh Bundle

通过 Chrome DevTools MCP 服务器驱动本地浏览器（Chrome / 360Chromex 等）进行网页调试、浏览器自动化、性能分析与网络检查的中文本地化技能。服务器以全局方式安装（npm install -g，位于 $(npm root -g)），可任选 MCP 直连/中转或 CLI 兜底使用。激活关键词：Chrome DevTools、浏览器自动化、网页调试、页面快照(take_snapshot)、元素交互(点击/填写/拖拽)、性能分析(Lighthouse/Performance Insight)、内存泄漏排查、网络请求检查、控制台日志、网页截图。适用场景：调试网页或 Web 应用、自动化点击/填写/导航、分析 LCP/内存/可访问性、抓取页面结构与控制台、连接已登录浏览器复用登录态。不适用场景：纯后端或 CLI 任务、无需浏览器的数据处理、本机无可用浏览器且未用 verify_browser 指定路径的情况。

zhangweildlh Bundle

纯本地、离线、只读的代码知识图谱（影响面分析）引擎：将本地已索引工作树构建为调用图/使用图/继承图，提供符号搜索、调用链追踪、影响面评估、死代码定位、本地 git 变动爆炸半径映射等 15 个 MCP 工具。关键词：代码知识图谱、调用链追踪、影响面分析、架构检索、本地索引。当用户要求分析/阅读/修改/重构本地代码、定位/修复 BUG、审计/审查代码、回应 PR 审查意见、接手代码/项目，且目标位于本地已索引工作树时触发；当用户说"理解结构/评估影响面/追踪调用链/定位死代码"时触发。适用于本地研发认知任务（理解结构、改前评估影响、改后验证半径）。不适用于纯新增代码（无既有图可查）、Write/Edit 与 git 写动作本体、未克隆的远程 GitHub 仓库浏览（走 gh + github-personal-manager）、运行时调试；经 dmcp 分组 codebase-memory-mcp 或原生 stdio 直连调用。

zhangweildlh Bundle

当需要与 mimo.code（小米 MiMo 代码智能体）协同开展任何工程任务时加载本技能——包括但不限于：代码编写、修改、BUG 修复、代码审核、项目分析与重构、技术方案/架构讨论与文档编写、非代码文件（配置/文档/规格）讨论与编写、GitHub 变更的内容生成与审阅。覆盖 mimo.code 的能力边界、环境无关的通用调用方法（同时适配「Dynamic-mcp 工具中转连接」与「MCP 服务方式直接连接」两种接入形态）、以及「主Agent × mimo.code 通用分工闭环 + 双范式（串行交替审核 / 并行交叉审核）+ 强制前置分析约束（子任务启动前先判定场景与范式）+ 防死锁」的协同工作流。即使完全未接触过 mimo.code，读完本技能也可正确调用并组织协同。

zhangweildlh Bundle

让 Agent 记录经验、偏好与修正结果，从错误中学习、提炼经验教训，并随着时间推移不断优化自身行为。当用户说出 "improve yourself"（改进你自己）、"learn from that mistake"（从那次错误中学习）、"log what went wrong"（记录出错之处）、"review your lessons"（回顾你的教训）、"run a self-audit"（运行自我审计）、"check your soul file"（检查你的灵魂文件）、"update your playbook"（更新你的行动手册）等指令时使用；或当智能体检测到自身犯了错误、应当予以记录时使用。此外，该功能还会在会话开始时触发，以加载既往的学习成果，并定期触发以检测反复出现的错误模式。