본문으로 건너뛰기
cccskills
무료GitHub에 공개

interactive-flowchart-builder

Build interactive "living workflow" flowcharts in the visual style of https://statuesque-starship-49757d.netlify.app/revenue-engine - vanilla HTML/CSS/JS files with a horizontal step bar, a vertical trunk that grows per step, tool-to-tool columns with animated payload pills between brand-marked tool nodes, alternating left/right step descriptions, and per-step outputs. Use whenever the user says "build a flowchart", "interactive flowchart", "living workflow", "revenue engine flowchart", "demand gen flowchart", "GTM workflow diagram", "process diagram with animations", "Nebor-style flowchart", or asks to recreate the workflow visual from the Nebor site. Always asks which flowchart to build (Revenue Engine / Demand Gen / CRM & RevOps / Custom) and waits for explicit approval of the step + tool + edge table before generating any files.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

연동 도구HubSpotNetlify
설치 방법 보기

포함된 파일(1)

  • SKILL.md8.2 KB

SKILL.md(원문)

설치하기 전에 에이전트에게 전달되는 지시 내용을 확인할 수 있습니다.

When to invoke

User asks for an interactive multi-step process flowchart with tools, edges, and animated data flow - especially one modeled on the Nebor site's "living workflow" pattern. If the user just wants a static diagram, a Mermaid chart, or a slide-deck flowchart, hand off to a different skill - this one specifically produces the Nebor-style vanilla-JS animated flowchart.

Reference visual

The output should match the rendering pattern of:

All three use the same vanilla-JS engine driven by two globals: window.NEBOR_TOOLS and window.NEBOR_WORKFLOW. Reproduce that engine.

Workflow

Step 1 - Ask which flowchart to build

Print exactly this block and wait:

Which flowchart would you like me to build?

  1. Revenue Engine / Sales Engine & GTM - Identify → Enrich → Personalize → Send → Qualify → Book
  2. Demand Gen & ABM - audience build → signal capture → orchestrated plays → measurement
  3. CRM & RevOps - audit → model → automate → report → govern
  4. Custom - tell me the domain, steps, and tools and I'll design it

Reply with 1, 2, 3, or 4 (with details).

Step 2 - Draft and confirm the scope

Once the user picks an option, produce a compact markdown table with:

  • 4-7 steps (verb-first names, e.g. "Identify", "Enrich")
  • One-line desc per step
  • 3-5 tools per step (real product names, real hex brand colors)
  • roles map (one sentence per tool, per step)
  • flow edges - each with from, to, proto (one of api / webhook / event / sql / llm), label
  • One-line output per step (e.g. "1,284 ranked accounts written to HubSpot - weekly refresh")

End with: "Approve as-is, or change anything before I generate the files?"

Do not generate any files until the user types an approval (yes / approved / ship it / etc.).

Step 3 - Generate the files

After approval, write three files into the current working directory (or a path the user specifies):

  1. index.html - DOM skeleton + inline CSS + script tags
  2. tools.js - populates window.NEBOR_TOOLS
  3. workflow.js - populates window.NEBOR_WORKFLOW + the renderer + animation loop

Required DOM skeleton (in index.html)

<section class="workflow">
  <div class="workflow-track"><div class="workflow-track-fill"></div></div>
  <div id="wf-nodes" class="wf-nodes"></div>
  <h3 id="wf-title"></h3>
  <p  id="wf-desc"></p>
  <div id="wf-cumulative" class="wf-cumulative"></div>
  <div id="wf-output" class="wf-output"></div>
</section>
<script src="tools.js"></script>
<script src="workflow.js"></script>

Required data shape (in tools.js + workflow.js)

window.NEBOR_TOOLS = {
  "<slug>": {
    name: "<Display Name>",
    color: "#HEX",
    cat: "<Category>",          // CRM, Data, Enrichment, LLM, Outreach, Analytics, etc.
    fb:  "Xx",                  // 2-letter fallback mark
    domain: "<example.com>",    // optional
    desc: "<one-sentence why-this-tool>"
  }
};

window.NEBOR_WORKFLOW = [
  {
    step:  "Identify",
    desc:  "TAM mapped from firmographics, tech-stack, and intent.",
    tools: ["apollo", "clay", "builtwith", "hubspot"],
    roles: {
      apollo: "Sources accounts from firmographic + funding filters",
      clay:   "Orchestrates enrichment passes across 50+ providers",
      // one role per tool
    },
    flow: [
      { from: "apollo",    to: "clay",    proto: "api",     label: "POST /accounts" },
      { from: "builtwith", to: "clay",    proto: "api",     label: "tech.signals" },
      { from: "clay",      to: "hubspot", proto: "webhook", label: "ranked.json" }
    ],
    output: "1,284 ranked accounts written to HubSpot - weekly refresh"
  }
];

Render rules

  1. Build a horizontal step bar from NEBOR_WORKFLOW. Auto-advance every 4-6s; clicking a step jumps to it.
  2. Below the bar, render a vertical "trunk" - one continuous line connecting every step's root and merge markers.
  3. Each step renders a .tree-section containing: .trunk-root (number + step name) → .trunk-fork (horizontal bar) → .trunk-cols[data-cols=N] (columns of from-tool → animated .tree-payload on .tree-wire → to-tool) → .trunk-merge.
  4. Alternate the step description side: even index → text right, odd → text left.
  5. Animate ALL visible payload pills in sync (not just the current step's) - the "living" quality is the defining feature.
  6. Past steps render with class past and status pill "wired"; the current step renders with class current and pill "live".
  7. Tool nodes show: brand-color square mark on the left (inline SVG if available, else the 2-letter fb mark) + tool name + hover tooltip with the step-specific role.
  8. Color-code .tree-payload by proto:
    • api - indigo (#4F46E5)
    • webhook - teal (#0D9488)
    • event - amber (#D97706)
    • sql - violet (#7C3AED)
    • llm - rose (#E11D48)

Style targets

  • Background: warm off-white #F5F1EA. Ink: #15140F.
  • Serif display for headings (Tiempos, Cormorant, or Playfair Display fallback). Grotesk for body (Inter / system-ui).
  • Tool node: rounded rectangle, white fill, 1px ink border, brand mark on the left.
  • Payload pill: small pill shape, color per protocol, animated translateX along the wire with CSS keyframes.
  • Magazine-grade whitespace. No drop shadows, no glassmorphism, no gradients except subtle.

Step 4 - Hand off

After writing files, print:

  • How to preview: cd <dir> && python3 -m http.server 8000 then open http://localhost:8000.
  • How to add a step: append an object to NEBOR_WORKFLOW in workflow.js.
  • How to add a tool: add a slug entry to NEBOR_TOOLS in tools.js.
  • Where to tweak protocol colors / animation speed (note exact CSS class names).

Anti-Patterns

  • Don't reach for React, ReactFlow, Mermaid, or D3. The reference engine is vanilla JS by design - introducing a framework defeats the point.
  • Don't generate files before the user approves the scope table. Skipping the confirmation step produces flowcharts the user has to redo.
  • Don't invent tools or hex colors. Use real product names and real brand colors. If you don't know a color, ask before guessing.
  • Don't drop the animation. A static flowchart fails the brief.
  • Don't merge tools.js and workflow.js into one file unless the user asks - keeping them split is what makes the flowchart editable.
  • Don't write a README or extra docs unless the user asks. Three files is the deliverable.

Quality Checklist

  • User explicitly approved the step + tool + edge table before any file was written.
  • 4-7 steps total, each with step, desc, tools, roles (one per tool), flow, output.
  • Every tool slug referenced in any flow.from, flow.to, or tools[] exists as a key in NEBOR_TOOLS.
  • Every edge has a valid proto (api | webhook | event | sql | llm) and a human label.
  • Horizontal step bar + vertical trunk + animated payloads all render from a single index.html open in the browser (no build step).
  • Past steps show as "wired", current step shows as "live".
  • Hover on a tool node shows the step-specific role tooltip.
  • Brand-color marks render for every tool (SVG or 2-letter fallback).

리뷰

아직 리뷰가 없습니다. 사용해 본 소감을 남겨 주세요.

댓글

로그인 없이 쓸 수 있습니다. 사용 후기, 질문, 작동하지 않은 환경 등을 편하게 남겨 주세요.

아직 댓글이 없습니다. 첫 댓글을 남겨 보세요.

보내면 바로 공개됩니다. 이메일 주소나 전화번호 같은 개인 정보는 쓰지 마세요. URL은 2개까지입니다.

0 / 1,000자

이 양식은 reCAPTCHA로 보호되며 Google 개인정보처리방침 및 서비스 약관이 적용됩니다.

같은 저장소의 스킬

Rate any product, feature, or experience on the 11-star scale (Brian Chesky's Airbnb thought experiment). Use when user says "rate this experience", "11-star", "star rating", "experience audit", "how good is this", "experience rating", "product audit", "quality assessment", or wants to evaluate product quality and identify improvement paths. Also trigger when user wants to benchmark a feature, assess where a product stands, or map out what "great" looks like - even if they don't explicitly say "11-star".

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

qa-aman/claude-skills212026년 10월 13일 업데이트

Generate A/B copy variants grounded in Eugene Schwartz's awareness and market-sophistication stages and John Caples' tested-headline discipline. Every variant declares its angle, the awareness stage it targets, and a falsifiable hypothesis, and the skill recommends which pair to test first and what delta counts as signal. Use when the user says "A/B test this", "write variants", "give me options", "test different angles", "copy variants", "ab copy", "test this headline", "multiple versions", "which version is better", or wants to test copy before publishing. For reviewing copy that already exists, see copy-review. For sample size and stopping rules, see growth-experiment. For diagnosing a page rather than its words, see page-cro. For full paid ad sets by platform, see ad-campaign-writer.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

qa-aman/claude-skills212026년 10월 13일 업데이트

Write measurable acceptance criteria for requirements or user stories. Use when the user says "write acceptance criteria", "definition of done", "how do we test this", "fit criteria", "given when then", "Gherkin scenarios", "what does done look like", "testable conditions", "how will we know this works" - even if they don't explicitly say "acceptance criteria".

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

qa-aman/claude-skills212026년 10월 13일 업데이트

Build a strategic account plan for winning, retaining, or expanding a named account. Use when the user says "build an account plan", "strategic account planning", "how do I grow this account", "account expansion strategy", "whitespace analysis", "QBR prep", "key account review", "map the stakeholders in this account", "land and expand plan", "how do I break into [company]", or wants a structured strategy for a specific account.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

qa-aman/claude-skills212026년 10월 13일 업데이트

Write paid ad copy for LinkedIn, Google, Meta, and YouTube. Produces multiple variants per platform, each targeting a different Eugene Schwartz awareness level and Cialdini persuasion principle. Use when the user asks for ad copy, ad variants, "write LinkedIn ads", Google ads, Meta ads, paid social copy, ad headlines, or wants creative for a paid campaign. Reads brand voice, ICP, and positioning from knowledge/. For organic social posts, see linkedin-post or social-calendar. For the landing page the ad points at, see landing-page-writer. For testing variants, see ab-copy-writer.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

qa-aman/claude-skills212026년 10월 13일 업데이트

Design clean and consistent APIs. Use when the user says "design an API", "API design review", "design these endpoints", "REST API for X", "GraphQL schema", "API contract", "how should this endpoint work", "review my API design", or wants to create or improve an API interface - even if they don't explicitly say "API design".

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

qa-aman/claude-skills212026년 10월 13일 업데이트

qa-aman의 스킬 모두 보기

이 스킬의 문제 신고