AI & ML

grasscaograss Skill

知识库领域级重组工具。支持健康检查（inspect）、领域合并/拆分/移动/合并碎片（merge/split/move/consolidate），agent 驱动文件归属判定与合并策略，所有操作后自动同步 INDEX.md 与交叉引用。当用户说"重组知识库"、"整理 knowledge"、"合并领域"、"拆分领域"时触发。

grasscaograss Skill

并行 Agent 开发工作流：先生成接口契约，再按文件归属并行实现，最后集成验收。触发场景：(1) 用户要求开发跨多层的功能（领域+应用+前端+测试）；(2) 用户说"并行开发""parallel feature""多 agent 同时做"；(3) 新功能涉及 3 个以上文件切面时主动建议使用。

mindgoblinstudios Bundle

Guided install of Grimoire's Tome: explain the install model, walk the user through cloning, copying, or symlinking the parts they want, and hand off to council onboarding. Manual invocation only.

mindgoblinstudios Bundle

(grim:dev:model-quirks): (Model Quirks): Apply model-specific prompt mitigations inline (unlabeled) into AGENTS.md, skills, and prompts, and record every applied change in the reverse-index ledger at docs/model-quirks-ledger.yaml so it can be verified, pruned, or undone when models rotate.

mindgoblinstudios Bundle

Three Minions in a Trench Coat: manual-only multi-model coding council that runs three top Minion routes in parallel, compares their answers, and synthesizes the best final answer.

skyosev Skill

Use when an LLM-facing document — a prompt, skill file, spec, or instruction set — is verbose and must fit a tighter token budget. Not for user-facing copy, legal text, changelogs, or quoted material.

skyosev Skill

Review a GitHub pull request — fetches context, builds an intent model, runs 3 parallel review agents (Correctness, Conventions, Efficiency & Safety), validates findings against actual code, and publishes one structured review. Triggers on "review this PR", "review PR #123", "review github.com/owner/repo/pull/N", "check this pull request". Reports omit empty sections — no placeholder headings, empty tables, or negative statements like "no issues found".

skyosev Bundle

Use when reviewing Go, Python, or TypeScript code for invariants left to runtime, convention, or type assertions: tightening a domain model after prototyping, reducing casts and type-system bypasses, reviewing optionality and discriminated unions, or establishing a safety baseline before a refactor. Defaults to the codebase. Covers unguarded type assertions, loose optionality, defensive access in non-boundary code, leaky discriminated unions, runtime checks promotable to types, type-system bypasses, and Go's unchecked errors, nil pointer risks, zero-value traps, error chain correctness, context misuse, panic/recover misuse, and race conditions.

skyosev Skill

Audit Go code for design principle violations — god packages, rigid extension points, broken interface contracts, fat interfaces, and concrete dependency chains. Adapted from SOLID for Go's composition-over-inheritance model. Use when: reviewing package structure, preparing for extension with new variants, reducing coupling between packages, or improving testability.

davie521 Skill

Multi-source deep research. Prefers the exa MCP for search (firecrawl for scraping), falls back to the bundled /deep-research workflow or built-in WebSearch when they are absent. Searches the web, synthesizes findings, and delivers cited reports with source attribution. Use when the user wants thorough research on any topic with evidence and citations.

davie521 Skill

Default code review route — runs Codex (read-only `codex exec` via the codex-run wrapper) with the multi-language code-review methodology (severity matrix, file:line, scope triage, mandatory security pass). Read-only, no edits. Trigger when the user says "code review" / "审一下" / "review 一下" / "审这个 diff" / "用 codex 审" / "second opinion" / "复审" / "看看这次改动" or asks for any code review on a diff, file, or PR. This is the preferred entry point — route to the built-in /code-review skill when the user wants inline PR comments (--comment) or the findings auto-applied (--fix), and to the Claude-subagent dispatcher (code-review:code-review) only when the user explicitly asks for "Claude reviewers" / "subagent review" / "the multi-agent code-review".

davie521 Bundle

Triage issues through a state machine driven by triage roles. Use when user wants to create an issue, triage issues, review incoming bugs or feature requests, prepare issues for an AFK agent, or manage issue workflow.

davie521 Bundle

Visualize whether skills, rules, and agent definitions are actually followed — auto-generates scenarios at 3 prompt strictness levels, runs agents, classifies behavioral sequences, and reports compliance rates with full tool call timelines

creatorland Skill

Guided first run of the Creatorland Data MCP — one live call of each of the three tools with safe example arguments, narrated so a new workspace learns the tools, the credit model, and the privacy floors in ~5 minutes for ~8 credits. Use when the user says "onboarding tour", "show me how this works", "get started with creatorland", "give me a tour", "what can this MCP do", or when this appears to be a first session in a new workspace with no prior Creatorland calls. Ends with the user's first real shortlist invitation.

creatorland Skill

Plan and track credit spend across a session before running expensive MCP workflows — estimate cost up front, hold a per-job ledger, warn over a threshold, and produce a chargeback log. Use when the user says "how many credits will this cost", "track my credit spend", "set a credit budget", "don't go over N credits", or "log credits for chargeback". Meta-skill, 0 data credits — it does no market/profile calls itself, it only prices and accounts for other skills.

imtangyujing Bundle

飞书视频会议：搜索历史会议、查询会议纪要产物（总结、待办、章节、逐字稿）、查询会议参会人快照。1. 查询已经结束的会议数量或详情时使用本技能（如历史日期｜昨天｜上周｜今天已经开过的会议等场景），查询未开始的会议日程使用 lark-calendar 技能。2. 支持通过关键词、时间范围、组织者、参与者、会议室等筛选条件搜索会议。3. 获取或整理会议纪要、逐字稿、录制产物时使用本技能。4. 查询“谁参加过某会议”“参会人列表”等参会人快照信息用 vc meeting get --with-participants（任意时点可查，含已结束会议）。注意：**Agent 真实入会/离会、感知正在进行中会议的实时事件**请使用 lark-vc-agent 技能，本技能不覆盖写操作和会中事件流。

imtangyujing Bundle

飞书事件订阅：通过 WebSocket 长连接实时监听飞书事件（消息、通讯录变更、日历变更等），输出 NDJSON 到 stdout，支持 compact Agent 友好格式、正则路由、文件输出。当用户需要实时监听飞书事件、构建事件驱动管道时使用。

imtangyujing Bundle

Turn one or many articles, web pages, video/audio links, local recordings, platform transcripts, event transcripts, and raw interview texts into usable Markdown context in the resolved output language for writing and Obsidian. Use when the user asks to fetch articles, transcribe podcasts or videos, connect audio/video material to a text workflow, extract notes, polish transcript prose, or process multiple source links in parallel. Keep one source owner responsible for each source from acquisition through final Markdown; use the current main agent for one source and one end-to-end worker per source for batches.

imtangyujing Bundle

飞书视频会议：让机器人代当前用户加入/离开正在进行的会议，并读取会议期间的实时事件（参会人加入与离开、发言、聊天、屏幕共享等）。1. 用户提供 9 位会议号、要求代为入会或离会时使用 +meeting-join / +meeting-leave——会真实产生入会/离会记录。2. 会议进行中用户想知道“谁加入了”“谁离开了”“谁在发言”“有人共享屏幕吗”等会中动态时，机器人入会后用 +meeting-events 读取事件时间线。3. 典型场景：参会机器人、会中助手、代为旁听、代为参会。前提：机器人只能读到它自己参会过且仍在进行中的会议的事件；查询已结束会议的参会名单、纪要或逐字稿请使用 lark-vc 技能。

ulpi-io Bundle

MANDATORY for starting any conversation or task. Establishes required workflows: discovering and using skills, invoking the Skill tool before announcing usage, performing brainstorming before coding, creating TodoWrite todos for all checklists, and selecting/invoking the correct specialized agent persona for the task domain.

ulpi-io Skill

Hybrid vector+BM25 code search, symbol lookup, dependency analysis, PageRank importance scoring, coupling metrics, and circular dependency detection via the codemap CLI. Use when the user asks to search code, find symbols, analyze dependencies, find dependents, check coupling, view PageRank importance, detect circular dependencies, index a project, get file summaries, or explore architecture health. Wraps the @ulpi/codemap CLI (not MCP tools).

ulpi-io Bundle

Laravel 12 API development — API-only, Actions pattern, Sanctum/Passport auth, Eloquent strict mode, Redis caching/queues, Horizon, Filament admin, Pest testing. Also covers Laravel AI ecosystem: AI SDK (agents, embeddings, image/audio generation), Boost (AI-assisted development), and MCP (exposing your app to AI clients). Use when working on any Laravel route, controller, model, action, migration, job, notification, test, API endpoint, AI agent, or MCP server task.

ulpi-io Skill

Credential management for AI coding agents — encrypted vault, CLI injection, MCP shim

ulpi-io Skill

Find, install, and manage AI agents across 43+ coding IDEs. Search the agentshq registry, install agents from GitHub repos or URLs, list/remove/update installed agents. Invoke via /find-agents or when user says "find an agent", "install agent", "search agents", "list agents", "remove agent", "update agents".

ulpi-io Skill

Spawn a separate Claude Code agent to review code changes made in the current session or branch. Uses the Agent tool with worktree isolation so the reviewer gets an independent read-only copy. Use when the user says "claude review", "self-review", "/claude-review", or when a task plan calls for post-task review.

ulpi-io Skill

Extract learnings from a session and propagate them to appropriate agent files based on scope (Global for all subagents, Claude Code Only for main agent, or Agent-Specific). Use when a session revealed patterns, mistakes, or insights. Invoke via /update-agent-learnings or after challenging sessions.

ulpi-io Bundle

Detect tech stack, generate Claude Code settings.local.json permissions and MCP server suggestions. Analyzes lock files, config files, frameworks, services, and monorepo structure. Invoke via /update-claude-settings or when user says "audit settings", "setup permissions", "configure claude settings".

ulpi-io Bundle

Extract learnings about Claude Code behavior from a session and add them to the project's CLAUDE.md file. Use for behavioral rules, workflow patterns, and project-specific instructions that apply to the main Claude Code agent. Invoke via /update-claude-learnings after sessions revealing behavioral patterns.

ulpi-io Bundle

Orchestrate multiple specialized agents working in parallel when building independent features. Use when the task list contains 3+ unrelated tasks that don't share state or files. Matches each task to the right expert agent and launches them concurrently via the Agent tool with worktree isolation. Supports all available subagent types.

ulpi-io Bundle

Orchestrate multiple specialized agents working in parallel to debug independent problems. Use when encountering 3+ unrelated bugs or test failures in isolated modules. Matches each problem to the right expert agent and launches them concurrently via the Agent tool with worktree isolation. Supports all available subagent types.

nemori-ai Bundle

cc-master 仓库自用的 git worktree 隔离纪律——Solo（单任务单 worktree）与 Fan-out（多 worktree 并行）两种操作模式 + 一组防止并行开发污染 main checkout 或彼此的红线。何时用 / Use when：(1) 开始一段需要与 main checkout 隔离的实现工作、动第一次文件编辑之前；(2) 把并行实现 fan-out 到多个 worktree；(3) 决定这场并行怎么收口（各自 feature-branch PR vs HUB 集成分支合并回）；(4) 一次 merge / PR 落地后的清理墙、开下一个 PR 之前；(5) 给一个将在 worktree 内工作的 sub-agent 写派发 prompt；(6) 作为 master orchestrator dogfood 本仓、fan-out 后台 worker 之前先立起隔离拓扑。Triggers: worktree、隔离、fan-out、并行改文件、cd 进哪个树、main 被污染、清理残留 worktree、spoke 该不该 commit。Do NOT use when：把目标切成任务 DAG / 排 wave（那是 slicing-goals-into-dags + master-orchestrator-guide 的 decomposition），worktree 内的 TDD / 测量循环（dev-as-ml-loop / engineering-with-craft），或 PR 创建与 merge 机制（AGENTS.md §11 的 gh 手工流——本仓没有 github-pr skill）。

nemori-ai Bundle

当你创建、编辑或审查一个 cc-master skill 时用——尤其是那种纪律型（discipline-enforcing）skill（master-orchestrator-guide、authoring-workflows、本 skill），它的规则 agent 在压力下能合理化掉。Triggers: 新建/修改/审查本仓 skill、加 Rationalization Table / Red Flags / 决策程序、改 SKILL.md 的纪律段或 description；或当你发现自己正打算先动手写 skill 正文、却没先看一个 agent 失败。Do NOT use when 你只是要判断要不要建一个 skill / 这块该 skill 还是 reference / 两个 skill 重不重叠（那是 curating-skill-portfolios）；Do NOT use when 你要声明 J / 跑触发或行为 eval / 度量一个 skill（那是 grounding-skill-evals）；Do NOT use when 你要跑 skill knowledge graph 正式规范/CLI（→ design_docs/skill-knowledge-graph）。

nemori-ai Bundle

Use when 设计、审查或重构 cc-master / 类似项目的多 agent harness plugin adapter：新增或修复 host adapter、建立 N-host capability parity、判断 Track A / Track B、为跨 hooks/commands/skills/ccm 的能力建立 Capability INTENT / hook CONTRACT，或处理 Cursor IDE plugin 的非 1:1 语义。覆盖 source-to-adapter、SAP/PHIP、host-native manifest / path / hook / command 边界。Do NOT use when 只实现投影脚本（adapter-projection-engineering）、只做打包发布（plugin-release-engineering）、只改一个 skill body（cc-master-skillsmith）、只跑 skill knowledge graph 正式规范/CLI，或只设计不改 plugin adapter 的 cross-harness headless CLI worker transport；后者以 design_docs/cross-harness-orchestration-capability-model.md 为合同 SSOT，按普通 engineering / dev loop 推进。

nemori-ai Bundle

Use when dev work should be run as an ML-style optimization loop, either by a master orchestrator shaping dev-task handoffs/subagent roles or by an execution agent driving one task to acceptance —— 当你要把开发工作当成优化过程来跑:master orchestrator 用它设计外层 dev loop/objective/measurement/subagent 组件分工/restart-stop/board ledger 语义,执行 agent 用它把单个开发任务推进到验收。心智锚:验收=目标函数、loop=迭代优化(提议→测量→调整)、测试=测量仪器/梯度、explore vs exploit、局部最小值=钻牛角尖→restart、收敛即停别过拟合、拟合意图非用例、简单性=正则、持续用 board 维护优化状态以便 compact 后续接。Triggers: 派发或接手 dev 任务、要把 work order 改造成可测优化问题、要设计 subagent 分工/测量/验收/重启条件、要用 board 管理优化目标和迭代状态、compact 后续接 dev loop、卡住在一个方案上越改越深(钻牛角尖)、怎么判断"做完了"、要不要先写测试/测量、要不要换方案。Do NOT use when 你在决定顶层该编排什么 / WIP / 临界路径 / HITL / 配额(master-orchestrator-guide)、怎么把目标切成任务 DAG(slicing-goals-into-dags)、领域 / 类 / 合约 / 测试本身怎么建得好——DDD/OOP/SDD/TDD 手艺内容(engineering-with-craft)、怎么用 ccm 写 board(using-ccm)、workflow 脚本怎么写(authoring-workflows)。

hack23 Skill

Citizen Intelligence Agency data model integration, riksdag data processing, political entity mapping

hack23 Skill

AI-enabled security excellence through transparent ISMS implementation, defense-in-depth, and strategic planning aligned with Hack23 business model