Skip to content

Presentation — HTML tab-bar deck from PRD / Spec / ADR / Plan

Input doc: $ARGUMENTS. If empty, ask for the path (defaults live in docs/prd/, docs/specs/, docs/adr/, docs/plans/).

You receive ONE markdown file — a PRD (Product Requirements Document), a Feature Specification, an ADR (Architecture Decision Record), or an implementation Plan — and render it into a single self-contained .html presentation that opens directly in a browser (no build step, no VS Code extension).

0. Golden rules

  • Input is exactly one .md file. If none is given, ask for the path. Good defaults: docs/prd/PRD-*.md, docs/specs/spec-*.md, docs/adr/0xxx-*.md, docs/plans/plan-*.md.
  • Output is exactly one .html file. Render only — never add, drop, or reinterpret content. Keep wording, decisions, tables, code blocks, diagrams, and estimates faithful. You may compress prose into bullets, but no new facts.
  • Reuse the template's classes only. Build from template-tabbar.html in this skill folder. Do not invent new CSS.
  • Do not open the browser / run Start-Process unless the user asks.

1. Workflow

  1. Read the source .md and detect its type from the title / ## Metadata / heading shape:
    • # PRD: or ## Problem Statement + ## Goals & Success Metrics + ## User Stories → PRD
    • # Feature Specification: or ## Functional Requirements → Spec
    • # ADR- / ## Decision Drivers / ## Considered Options → ADR
    • # Plan: / ## Implementation Steps / ### Phase N → Plan
    • If ambiguous, pick the closest mapping below; don't ask unless truly unclear.
  2. Copy template-tabbar.html to the output path (see §3).
  3. Fill the <header class="hero">: eyebrow = type + id (e.g. ADR-0006), <h1> = document title, chips = Status / Date / Owner / Version pulled from ## Metadata when present. Keep the .back-link (← Presentations) anchor from the template — it links to https://nguyenhuuca.github.io/assessment/presentations/.
  4. Map sections to tabs — one .tab-btn + matching .panel per major ## section, using the type-specific maps below. First tab gets active / first panel gets show.
  5. Fill panels with cards. One .card per logical unit. Use accent colors to encode meaning (see §2). Put long rationale, alternatives, and code inside <details>. Keep tables as <table>, ASCII diagrams and code as <pre>.
  6. Verify the .html exists and every data-tab has a matching panel id.
  7. Summarize for the user: output location + navigation (← / → to switch sections, click 🔎 to expand details).

2. Color & block conventions (consistent across all types)

Meaning Block
Goal / overview / primary path .card.blue, .hl.goal
Chosen / done / success / deliverable .card.green, .hl.prod
Requirement / rule / question / quote .card.cyan or .prompt
Caution / trade-off / consequence .card.amber
Risk / rejected option / breaking change .card.red
Data model / API / technical detail .card.violet
Note / example .card.yellow, .ex
Estimate / duration .time on the title (e.g. 1.5d)
Tag (FR-1, Phase 3, Decision) .tag

3. Output location

Default: write <doc-name>.html next to the source .md, same base name, only the extension changes (e.g. docs/adr/0006-hls.md → docs/adr/0006-hls.html). If the user prefers a dedicated folder, use docs/presentations/<doc-name>.html.

4. Type-specific tab maps

🎯 PRD → tabs

🎯 Overview · ❓ Problem · 📈 Goals & Metrics · 👤 User Stories · ✅ Requirements · 🧭 Scope · 🔗 Dependencies · ⚠️ Risks (include only sections present; fold Open Questions / Appendix / Approval / Version History into the nearest panel or a final 📌 Notes tab).

  • Overview / Problem Statement → .card.blue + .hl.goal.
  • Goals & Success Metrics → metrics in a <table>; goals as green bullets.
  • User Stories → one .card.cyan per story (or a list), role/need/value.
  • Requirements → .card.cyan cards; tag must-have vs nice-to-have with .tag.
  • Scope → In-Scope (green bullets) vs Out-of-Scope (muted) in one card.
  • Risks & Mitigations → <table> (risk, impact, mitigation) in .card.amber.

📐 Spec → tabs

🎯 Overview · 📐 Business Rules · ✅ Requirements (FR-*) · 🔌 API Changes · 🗄️ Data / DB · 🔒 Security · 🧪 Testing (include only sections present).

  • Each FR-N → one .card.cyan with a .tag (FR-1) and its acceptance criteria as a bullet list. Group several FRs per panel if short.
  • Each Rule → a row in a card or a .prompt.
  • API endpoints → a <table> (method, path, auth, purpose).
  • DB changes → <pre> for the DDL / table sketch in a .card.violet.

⚖️ ADR → tabs

📌 Context · 🎚️ Decision Drivers · ⚖️ Options · ✅ Decision · 📊 Consequences · 🔗 Links (and 📜 Changelog only if useful).

  • Considered Options → one .card per option; the chosen one is .card.green with a Chosen .tag, rejected ones .card.red/.amber. Pros/cons as sub-lists; put deep analysis in <details>.
  • Decision Outcome / Quantified Impact → .card.green + .hl.prod, metrics in a <table>.
  • Consequences → split positive (.card.green) vs negative (.card.amber).

🪜 Plan → tabs

🎯 Objective · 🧭 Scope · 🏗️ Technical Approach · 🪜 Phases · 📁 Files · ⚠️ Risks / Dependencies (include only what exists).

  • Scope → one card with In-Scope (green bullets) and Out-of-Scope (muted).
  • Implementation Steps / Phase N → each phase a .card, phase number as .tag, the time estimate as .time (e.g. 1.5d). Long phase detail/code in <details>. If there are many phases, one panel listing all phases as cards.
  • Files to Create / Modify → a <table> (path, action, note).

5. Handoff

After rendering, leave <doc-name>.html at the output path. Tell the user:

  • Open: double-click the .html (renders in any browser, no server).
  • Navigate: click a tab, or press ← / →; click 🔎 to expand details.
  • Update: when the source .md changes, re-run this skill to regenerate.

Pipeline: scope (PRD) → spec (Spec) / architect (ADR) / swarm-plan (Plan) → presentation (export .html deck). Any stage's artifact — PRD, Spec, ADR, or Plan — can be fed straight into this skill.