creative-toolkit

Files (7)

• SKILL.md
• copywriting-guide.md · references/
• creative-patterns.md · references/
• platform-cards-extraction.md · references/
• scripting-voice-extraction.md · references/
• toolkit-template.html · references/
• voice-dna-extraction.md · references/

SKILL.md (first lines)

# Creative Toolkit Builder

You produce a creative toolkit — a working library a creative team opens and starts producing
from immediately. The toolkit has two files: a CSV (100-150+ rows) and a companion HTML browser.

## The Output: 9 Sections Across Two Files

Every toolkit contains exactly these 9 sections. This is the structure of both the CSV and
the HTML. Each section serves a different person on the creative team.

### Section 1: Voice DNA Card (8-10 items)
The brand's extracted voice profile — archetype, voice attributes with "NOT" contrasts, 8
tone dimension positions, sounds-like / doesn't-sound-like examples. This is the taste
reference the team holds everything against.

### Section 2: Hook Bank (25-30 hooks)
Verbatim opening lines organized by psychological mechanism: Curiosity Gap, Contrarian
Premise, Pain Point Mirror, Social Proof, Urgency/Scarcity, Identity. Each hook is the
actual words a creator would say or a caption would start with. "Your morning bar has 14g
of sugar. Ours has 2." — that's a hook. "Use a curiosity gap hook" — that's a description,
not a hook. Every item in this section is ready to use as-is.

### Section 3: Angle Library (10-15 angles)
Strategic content framings with 2-3 sentence briefs. Each tells the creative team: what the
framing is, who it targets, and the data that proves it works.

### Section 4: Copy Bank (50+ lines)
Ready-to-use copy organized by where it goes: Caption Openers, Text Overlays, CTAs, Comment
Replies, One-Liners. Volume matters — 50 minimum, ideally 60-80. Mix of short punchy, medium,
and longer narrative lines. Every line contains at least one specific detail (a number, a time,
a name, a sensory detail).

### Section 5: Tagline Options (10-15 phrases)
Memorable phrases the brand could own, across registers: Bold, Playful, Authoritative, Minimal.
Each passes the swap test — if you could put a competitor's name on it, it's not specific enough.

### Section 6: Concept Briefs (5-8 concepts)
Full creative concepts a team could brief from. Each includes: concept name, format, hook
approach, beat-by-beat structure, platform, production notes, data proof, inspiration source.

### Section 7: Script Starters (3-5 scripts)
Beat-by-beat scripts with timestamp, visual direction, dialogue, text overlay. Enough to
film from. Matches the brand's scripting voice.

### Section 8: Visual / Format Direction (5-8 items)
Production guidance: editing pace, shot types, text treatment, music/audio, color/mood,
thumbnail/first-frame direction.

### Section 9: Inspiration Sources (8-12 items)
The actual posts, threads, and data points that inspired the toolkit. Organized by source
type with platform, description, and data proof.

## The CSV File

Columns: section, id, type, item, inspiration_source, source_detail, data_proof,
pattern_name, platform, audience, notes, mechanism (hooks), register (taglines), usage (copy).

100-150+ rows. This is the working file a creative team imports into their planning tool.

## The HTML File

Read `references/toolkit-template.html` before writing any HTML. The template defines the
exact design system:

- **Fonts:** Space Grotesk (headings), DM Sans (body), JetBrains Mono (data/metrics)
- **Background:** Cream (#f0ead6)
- **Primary cards:** Forest green (#1b4332)
- **Accents:** Lime (#c8e64a)
- **Highlights:** Yellow (#d4d069)
- **Item backgrounds:** White (#ffffff)
- **Feel:** Warm editorial — generous whitespace, padded cards, clean typography

The HTML has a sticky nav bar with section navigation. Each section is a visual block with
card types appropriate to its content (hook cards, copy line cards, concept brief grids, etc.).

## Every Toolkit Item Answers Four Questions

1. **What is it?** — The hook, angle, copy line, concept, or script
2. **Why does it belong here?** — Data proof (engagement multiples, outlier flags)
3. **Where did it come from?** — Inspiration source tag:
   - **Whitespace** — Nobody in the category is doing this
   - **Trend** — This pattern is gaining momentum right now
   - **Your Top Performer** — Iteration on something already working for this brand
   - **Category Performer** — A competitor is proving this works
   - **Audience Gap** — Real people are asking for this and no brand addresses it
4. **Does it sound like us?** — Written in the brand's extracted voice DNA

## Reference Files

Read these before building:

- `references/copywriting-guide.md` — **Read first.** AI slop blacklist, specificity rules,
  rhythm techniques, copywriting frameworks, voice adaptation checklist. Every piece of copy
  passes through this guide.
- `references/creative-patterns.md` — 133 named creative patterns. Your vocabulary for naming
  what you find in the data.
- `references/voice-dna-extraction.md` — Full voice DNA schema: archetypes, tone dimensions,
  language rules, scripting voice clusters.
- `references/scripting-voice-extraction.md` — Scripting voice clusters by content format.
- `references/platform-cards-extraction.md` — Per-platform voice adaptation cards.
- `references/toolkit-template.html` — The HTML design template. Read before building HTML.

## Workflow

### Step 1: Understand the Brief

Before pulling any data, clarify: the brand, the campaign/goal, the audience, 2-4 competitors,
target platforms, and any existing top performers or brand guidelines.

### Step 2: Get the Scope Right

Every read is about a **project** — a tracked set of sources inside a **portfolio**. Orient
first, then work in a project whose scope covers the focal brand, 2-4 competitors or creators,
and the communities the audience talks in.

1. `whoami` → `list_portfolios` → `list_projects({ portfolioId })`
2. Reuse a project, or `create_project({ portfolioId, name })` for this campaign
3. `get_project({ projectId })` — a project *is* its scope, so read what it currently covers
   before you trust a result. An empty-scope project reads the whole portfolio universe.

From adologyai/content-intelligence-plugin, Apache-2.0.

More from adologyai/content-intelligence-plugin