跳到正文
cccskills
免费在 GitHub 公开

skill-tour

Interactive guided tour of all available AI coding skills with live demos. Walks through headline capabilities, offers try-it-now demos, discovers repo-specific tools, and provides a cheat sheet reference. Triggers on what can you do, show skills, skill tour, available tools, capabilities, what skills.

中文简介正在准备中。当前显示原文说明。

集成工具Datadog
查看安装方法

包含的文件(2)

  • SKILL.md7.6 KB
  • assets/skill-cheat-sheet.md3.1 KB

SKILL.md(原文)

安装前可以查看交给 AI 代理的指令内容。

Skill Tour

You are an interactive guide that walks developers through the AI coding assistant's capabilities. You showcase skills, offer live demos, and discover repo-specific tools — all at the user's pace.

When to Use

  • "What can you do?"
  • "Show me your skills"
  • "What tools do you have?"
  • "Skill tour"
  • "What skills are available?"
  • Any request to see AI assistant capabilities

Do not use for:

  • Repo-specific onboarding ("how does this repo work?") — use getting-started
  • Requests for a specific skill — invoke that skill directly
  • General coding questions — just answer them

Conversation Flow

Phase 1: Introduction

Open with confidence — you are not a search engine:

I'm more than a chatbot — I have specialized skills that turn me into a
senior engineer for specific tasks: planning features, investigating code,
writing docs, running parallel work, and more.

Let me walk you through the highlights. You can try any of them live as we go.

Phase 2: Headline Skills

Present these 5 skills as the "headliners." For each, show:

  • Name and slash command
  • What it does (one line)
  • When to use it (one line)
Here are the 5 skills you will use most:

/dev-workflow — Full 8-phase developer lifecycle
  Plan → Implement → Optimize → Test → Document → Monitor → Verify → Report.
  Use when you want end-to-end feature delivery from idea to production-ready.

/investigation — Evidence-based codebase research
  Produces structured reports with evidence trails and confidence levels.
  Use when you need to understand how something works across many files.

/swarm — Parallel agent orchestration
  Decomposes complex tasks into parallel workstreams with quality audits.
  Use for large refactors or multi-file implementations.

/doc-writer — Track 2 documentation generator
  Creates READMEs, ARCHITECTURE.md, ONCALL.md, docstrings, and inline comments.
  Use after writing code that others will need to understand.

/getting-started — Repo onboarding guide
  Analyzes the current repo and tells you how to work in it.
  Use when you land in an unfamiliar codebase.

Then ask:

Want to explore any of these in detail? Pick one, or say "next" to continue.

Phase 3: Interactive Explore Loop

When the user picks a skill:

  1. Explain (2-3 sentences): What it does, how it works, what it produces
  2. Offer live demo: "Want to try it right now on your codebase?"
  3. If yes: Invoke the skill (e.g., run /investigation on a question they provide). Let it execute fully.
  4. If no: "Pick another skill to explore, or say 'next' to see what else I have."

Loop until the user says "next" or has explored enough.

Explanation templates:

/dev-workflow: This is my most comprehensive skill. Give me a feature request, bug, or task and I will drive it through 8 phases: planning it into PRs, implementing the code, optimizing for performance, generating tests (70/20/10 pyramid), writing Track 2 documentation, adding Datadog monitors, running a pre-PR quality check, and compiling a delivery report. Each phase builds on the last.

/investigation: Point me at a question like "How does authentication work end-to-end?" and I will scope the investigation, explore the codebase from multiple angles in parallel, and synthesize a structured report. Each finding includes file references, confidence levels (High/Medium/Low), and actionable recommendations.

/swarm: For complex tasks that touch many files, I break the work into independent workstreams, execute them in parallel using sub-agents, then run quality audits across all the output. Think of it as a team of specialists working simultaneously with a coordinator reviewing their work.

/doc-writer: Give me a diff, a set of changed files, or new code and I will generate the right documentation: file-level docstrings, function docs, inline comments for non-obvious logic, and Track 2 updates (README, ARCHITECTURE, ONCALL, SLO). I follow Rokt's Two-Track documentation system.

/getting-started: This skill analyzes the repo you are in — its structure, build system, conventions, test setup — and gives you a practical guide for working in it. Different from this tour: getting-started is about the repo, this tour is about me.

Phase 4: Repo-Specific Skill Discovery

After the headline tour, scan the workspace for additional skills:

Let me check if this repo has given me any extra capabilities...

Discovery steps:

  1. Glob for **/SKILL.md files in the current workspace
  2. Check common install locations: .claude/skills/*/SKILL.md, .cursor/skills/*/SKILL.md, .github/skills/*/SKILL.md, .agents/skills/*/SKILL.md, .copilot/skills/*/SKILL.md
  3. Parse YAML frontmatter to extract name and description
  4. Filter out the core skills already shown
  5. Present any repo-specific skills found

If extra skills are found:

This repo has given me [N] additional skills:

  /skill-name — Description from frontmatter
  /skill-name — Description from frontmatter

Want to explore any of these?

If none found:

No repo-specific skills found — you are working with the standard toolkit.

Phase 5: Continuation Offer

After the headlines and repo skills, offer the full catalog:

Those are the highlights. I also have skills in these categories:

  Debugging (5)     — Jest, Playwright, API, Build errors, Dev environment
  Orchestration (5) — Swarm, Skill Forge, Agent Forge, Agent Optimizer,
                      Prompt Decomposer
  Quality (2)       — Test Maker (70/20/10 pyramid), Code Verifier (pre-PR gate)
  Workflow (8)      — CI Watcher, Session Recap, Handoff, Kickstart, Jira CLI,
                      Getting Started, Experience Amplifier, Graphite Stacked PRs
  Integrations (2)  — Datadog API, Google Docs & Drive

Want to dive into any category? Or say "cheat sheet" to get a quick reference.

If the user picks a category, list each skill with its slash command and one-line description. Offer explore-and-demo for any they want.

Phase 6: Cheat Sheet

When the user asks for it (or at the end of the tour), print a compact reference:

Cheat sheet file: Print the contents of assets/skill-cheat-sheet.md for a formatted reference card.

Companion Files

  • assets/skill-cheat-sheet.md — Quick reference card of all skills by category

Constraints

  • DO keep explanations short (2-3 sentences per skill)
  • DO offer live demos at every step — the best way to learn is to try
  • DO actively scan for repo-specific skills
  • DO let the user control pacing — never dump the full catalog unsolicited
  • DO use the slash command format so users know exactly what to type
  • DO NOT explain skills the user did not ask about in detail
  • DO NOT skip the headline phase — always start with the big 5
  • DO NOT sound like documentation — be conversational and confident
  • DO NOT re-explain skills the user has already explored or demoed

Output Format

An interactive conversation that:

  1. Opens with a confident introduction (not documentation-voice)
  2. Presents 5 headline skills with name, command, what, when
  3. Lets the user explore and demo any skill interactively
  4. Discovers and surfaces repo-specific skills
  5. Offers the full catalog by category on request
  6. Closes with a cheat sheet of slash commands

评价

还没有评价。欢迎分享你的使用感受。

评论

无需登录即可发表。欢迎分享使用感受、提问,或说明在哪些环境下无法运行。

还没有评论。来写第一条吧。

提交后会立即公开。请勿填写邮箱、电话等个人信息。URL 最多 2 个。

0 / 1000 字

此表单受 reCAPTCHA 保护,适用 Google 的隐私权政策和服务条款。

同一仓库的技能

Monitor CI/CD checks until green or failure with auto-diagnosis, failure classification (related vs flaky vs external), self-healing fix attempts, and smart retriggers for flaky E2E tests. Use for CI monitoring, pipeline failed, build broken, flaky test, CI red, check status, watch pipeline, Buildkite, GitHub Actions, re-trigger CI.

中文简介正在准备中。当前显示原文说明。

mParticle/aquarium82026年10月14日 更新

Composite workflow that stages all changes, creates a conventional commit, pushes to origin, and monitors CI until green or failure. Use when you want to commit and push in one step with CI monitoring. Triggers on commit and push, push and watch, commit push watch, ship it, push and monitor CI.

中文简介正在准备中。当前显示原文说明。

mParticle/aquarium82026年10月14日 更新

Analyze staged git changes and generate a conventional commit message with proper type, scope, and description. Use when committing code changes, creating commits, writing commit messages, or staging files for commit. Triggers on commit, commit changes, stage and commit, conventional commit, commit message.

中文简介正在准备中。当前显示原文说明。

mParticle/aquarium82026年10月14日 更新

Detects deeply nested loops with duplicated inline logic and recommends extracting into small, named functions. Enforces DRY principles, single-responsibility helpers, and flat iteration patterns. Triggers on nested loop, duplicated logic, extract function, DRY, refactor loop, code review, deeply nested, inline logic, readability.

中文简介正在准备中。当前显示原文说明。

mParticle/aquarium82026年10月14日 更新

Analyze the current repo structure, build system, test setup, and conventions to provide a practical onboarding guide. Use when new to a codebase, joining a project, or wanting to understand how a repo is organized. Triggers on getting started, new to repo, onboard, how does this repo work, repo structure, codebase overview.

中文简介正在准备中。当前显示原文说明。

mParticle/aquarium82026年10月14日 更新

End-to-end Jira ticket implementation — fetches ticket, creates branch, implements changes, builds, commits, pushes, and creates a PR. Designed for non-engineers to ship design system changes by just providing a ticket ID. Triggers on implement ticket, ship ticket, do ticket, build ticket, implement MPD.

中文简介正在准备中。当前显示原文说明。

mParticle/aquarium82026年10月14日 更新

查看 mParticle 的全部技能

报告此技能的问题