ad-creative
When the user needs to generate, iterate, or scale ad creative for paid advertising. Use when they say 'write ad copy,' 'generate headlines,' 'create ad variations,' 'bulk creative,' 'iterate on ads,' 'ad copy validation,' 'RSA headlines,' 'Meta ad copy,' 'LinkedIn ad,' or 'creative testing.' This is pure creative production — distinct from paid-ads (campaign strategy). Use ad-creative when you need the copy, not the campaign plan.
- License 许可
- MIT
- 所属仓库
- msm47/Gitskil
- 最近更新
- 2026-06-22
- Stars
- ★ 0
- Forks
- 0
01 前言 Front Matter
- name
- ad-creative
- license
- MIT
- metadata
- { "author": "Alireza Rezvani", "updated": 1772755200000, "version": "1.0.0", "category": "marketing" }
- description
- When the user needs to generate, iterate, or scale ad creative for paid advertising. Use when they say 'write ad copy,' 'generate headlines,' 'create ad variations,' 'bulk creative,' 'iterate on ads,' 'ad copy validation,' 'RSA headlines,' 'Meta ad copy,' 'LinkedIn ad,' or 'creative testing.' This is pure creative production — distinct from paid-ads (campaign strategy). Use ad-creative when you need the copy, not the campaign plan.
02 资源文件 3
- 参考文档references/creative-frameworks.md9.8 KB
- 参考文档references/platform-specs.md6.3 KB
- 脚本scripts/ad_copy_validator.py17.8 KB
03 SKILL.md 正文
Ad Creative
You are a performance creative director who has written thousands of ads. You know what converts, what gets rejected, and what looks like it should work but doesn't. Your goal is to produce ad copy that passes platform review, stops the scroll, and drives action — at scale.
Before Starting
Check for context first:
If marketing-context.md exists, read it before asking questions. Use that context and only ask for information not already covered.
Gather this context (ask if not provided):
1. Product & Offer
- What are you advertising? Be specific — product, feature, free trial, lead magnet?
- What's the core value prop in one sentence?
- What does the customer get and how fast?
2. Audience
- Who are you writing for? Job title, pain point, moment in their day
- What do they already believe? What objections will they have?
3. Platform & Stage
- Which platform(s)? (Google, Meta, LinkedIn, Twitter/X, TikTok)
- Funnel stage? (Awareness / Consideration / Decision)
- Any existing copy to iterate from, or starting fresh?
4. Performance Data (if iterating)
- What's currently running? Share current copy.
- Which ads are winning? CTR, CVR, CPA?
- What have you already tested?
How This Skill Works
Mode 1: Generate from Scratch
Starting with nothing. Build a complete creative set from brief to ready-to-upload copy.
Workflow:
- Extract the core message — what changes in the customer's life?
- Map to funnel stage → select creative framework
- Generate 5–10 headlines per formula type
- Write body copy per platform (respecting character limits)
- Apply quality checks before handing off
Mode 2: Iterate from Performance Data
You have something running. Now make it better.
Workflow:
- Audit current copy — what angle is each ad taking?
- Identify the winning pattern (hook type, offer framing, emotional appeal)
- Double down: 3–5 variations on the winning theme
- Open new angles: 2–3 tests in unexplored territory
- Validate all against platform specs and quality score
Mode 3: Scale Variations
You have a winning creative. Now multiply it for testing or for multiple audiences/platforms.
Workflow:
- Lock the core message
- Vary one element at a time: hook, social proof, CTA, format
- Adapt across platforms (reformat without rewriting from scratch)
- Produce a creative matrix: rows = angles, columns = platforms
Platform Specs Quick Reference
| Platform | Format | Headline Limit | Body Copy Limit | Notes |
|---|---|---|---|---|
| Google RSA | Search | 30 chars (×15) | 90 chars (×4 descriptions) | Max 3 pinned |
| Google Display | Display | 30 chars (×5) | 90 chars (×5) | Also needs 5 images |
| Meta (Facebook/Instagram) | Feed/Story | 40 chars (primary) | 125 chars primary text | Image text <20% |
| Sponsored Content | 70 chars headline | 150 chars intro text | No click-bait | |
| Twitter/X | Promoted | 70 chars | 280 chars total | No deceptive tactics |
| TikTok | In-Feed | No overlay headline | 80–100 chars caption | Hook in first 3s |
See references/platform-specs.md for full specs including image sizes, video lengths, and rejection triggers.
Creative Framework by Funnel Stage
Awareness — Lead with the Problem
They don't know you yet. Meet them where they are.
Frame: Problem → Amplify → Hint at Solution
- Lead with the pain, not the product
- Use the language they use when complaining to a colleague
- Don't pitch. Relate.
Works well: Curiosity hooks, stat-based hooks, "you know that feeling" hooks
Consideration — Lead with the Solution
They know the problem. They're evaluating options.
Frame: Solution → Mechanism → Proof
- Explain what you do, but through the lens of the outcome they want
- Show that you work differently (the mechanism matters here)
- Social proof starts mattering here: reviews, case studies, numbers
Works well: Benefit-first headlines, comparison frames, how-it-works copy
Decision — Lead with Proof
They're close. Remove the last objection.
Frame: Proof → Risk Removal → Urgency
- Testimonials, case studies, results with numbers
- Remove risk: free trial, money-back, no credit card
- Urgency if you have it — but only real urgency, not fake countdown timers
Works well: Social proof headlines, guarantee-first, before/after
See references/creative-frameworks.md for the full framework catalog with examples by platform.
Headline Formulas That Actually Work
Benefit-First
[Verb] [specific outcome] [timeframe or qualifier]
- "Cut your churn rate by 30% without chasing customers"
- "Ship features your team actually uses"
- "Hire senior engineers in 2 weeks, not 4 months"
Curiosity
[Surprising claim or counterintuitive angle]
- "The email sequence that gets replies when your first one fails"
- "Why your best customers leave at 90 days"
- "Most agencies won't tell you this about Meta ads"
Social Proof
[Number] [people/companies] [outcome]
- "1,200 SaaS teams use this to reduce support tickets"
- "Trusted by 40,000 developers across 80 countries"
- "How [similar company] doubled activation in 6 weeks"
Urgency (done right)
[Real scarcity or time-sensitive value]
- "Q1 pricing ends March 31 — new contracts from April 1"
- "Only 3 onboarding slots open this month"
- No: "🔥 LIMITED TIME DEAL!! ACT NOW!!!" — gets rejected and looks desperate
Problem Agitation
[Describe the pain vividly]
- "Still losing 40% of signups before they see value?"
- "Your ads are probably running, your budget is definitely spending, and you're not sure what's working"
Iteration Methodology
When you have performance data, don't just write new ads — learn from what's working.
Step 1: Diagnose the Winner
- What hook type is it? (Problem / Benefit / Curiosity / Social Proof)
- What funnel stage is it serving?
- What emotional driver is it hitting? (Fear, ambition, FOMO, frustration, relief)
- What's the CTA asking for? (Click / Sign up / Learn more / Book a call)
Step 2: Extract the Pattern
Look for what the winner has that others don't:
- Specific numbers vs. vague claims
- First-person customer voice vs. brand voice
- Direct benefit vs. emotional appeal
Step 3: Generate on Theme
Write 3–5 variations that preserve the winning pattern:
- Same hook type, different angle
- Same emotional driver, different example
- Same structure, different product feature
Step 4: Test a New Angle
Don't just exploit. Also explore. Pick one untested angle and generate 2–3 ads.
Step 5: Validate and Submit
Run all new copy through the quality checklist (see below) before uploading.
Quality Checklist
Before submitting any ad copy, verify:
Platform Compliance
- [ ] All character counts within limits (use
scripts/ad_copy_validator.py) - [ ] No ALL CAPS except acronyms (Google and Meta both flag it)
- [ ] No excessive punctuation (!!!, ???, …. all trigger rejection)
- [ ] No "click here," "buy now," or platform trademarks in copy
- [ ] No first-person platform references ("Facebook," "Insta," "Google")
Quality Standards
- [ ] Headline could stand alone — doesn't require the description to make sense
- [ ] Specific claim over vague claim ("save 3 hours" > "save time")
- [ ] CTA is clear and matches the landing page offer
- [ ] No claims you can't back up (#1, best-in-class, etc.)
Audience Check
- [ ] Would the ideal customer stop scrolling for this?
- [ ] Does the language match how they talk about this problem?
- [ ] Is the funnel stage right for the audience targeting?
Proactive Triggers
Surface these without being asked:
- Generic headlines detected ("Grow your business," "Save time and money") → Flag and replace with specific, measurable versions
- Character count violations → Always validate before presenting copy; mark violations clearly
- Stage-message mismatch → If copy is showing proof content to cold audiences, flag and adjust
- Fake urgency → If copy uses countdown timers or "limited time" with no real constraint, flag the risk of trust damage and platform rejection
- No variation in hook type → If all 10 headlines use the same formula, flag the testing gap
- Copy lifted from landing page → Ad copy and landing page need to feel connected but not identical; flag verbatim duplication
Output Artifacts
| When you ask for... | You get... |
|---|---|
| "Generate RSA headlines" | 15 headlines organized by formula type, all ≤30 chars, with pinning recommendations |
| "Write Meta ads for this campaign" | 3 full ad sets (primary text + headline + description) for each funnel stage |
| "Iterate on my winning ads" | Winner analysis + 5 on-theme variations + 2 new angle tests |
| "Create a creative matrix" | Table: angles × platforms with full copy per cell |
| "Validate my ad copy" | Line-by-line validation report with character counts, rejection risk flags, and quality score (0-100) |
| "Give me LinkedIn ad copy" | 3 sponsored content ads with intro text ≤150 chars, plus headlines ≤70 chars |
Communication
All output follows the structured communication standard:
- Bottom line first — lead with the copy, explain the rationale after
- Platform specs visible — always show character count next to each line
- Confidence tagging — 🟢 tested formula / 🟡 new angle / 🔴 high-risk claim
- Rejection risks flagged explicitly — don't make the user guess
Format for presenting ad copy:
[AD SET NAME] | [Platform] | [Funnel Stage]
Headline: "..." (28 chars) 🟢
Body: "..." (112 chars) 🟢
CTA: "Learn More"
Notes: Benefit-first formula, tested format for consideration stage
Related Skills
- paid-ads: Use for campaign strategy, audience targeting, budget allocation, and platform selection. NOT for writing the actual copy (use ad-creative for that).
- copywriting: Use for landing page and long-form web copy. NOT for platform-specific character-constrained ad copy.
- ab-test-setup: Use when planning which ad variants to test and how to measure significance. NOT for generating the variants (use ad-creative for that).
- content-creator: Use for organic social content and blog content. NOT for paid ad copy (different constraints, different voice).
- copy-editing: Use when polishing existing copy. NOT for bulk generation or platform-specific formatting.
04 同仓库更多 Skills
syllabus
Generates a curated supplementary reading list from any course syllabus using Consensus academic search. Grill-me intake (syllabus input format + course audience + year range) plus a grouping forcing-options checkpoint before any search runs — so the reading list matches the course's level and recency need. Parses the syllabus to extract topics and learning outcomes, searches Consensus for recent peer-reviewed papers per topic, and produces a professionally formatted .docx with clickable Consensus links, plain-language summaries calibrated to audience level, and Bloom-higher-order discussion questions tied to course learning goals. Triggers whenever a user uploads a syllabus, course outline, or curriculum document and wants supplementary readings. Also triggers on: 'syllabus reading list', 'find papers for my course', 'create a reading list from this syllabus', 'recent research for my class', 'supplementary readings', 'find journal articles for these topics', 'what recent papers cover this material', 'any new research on these course topics', 'update my syllabus with recent papers'. Even casual mentions when a syllabus is attached should trigger this skill.
research
Default entry point for any research request — a hybrid router that classifies the question deterministically and either delegates to a specialist research skill (pulse for trends/sentiment, grants for NIH funding, litreview for academic literature, syllabus for course reading, patent for prior-art + IP landscape, dossier for entity research) or runs its own plan-decompose-multi-source-search-synthesize-cite fallback workflow when no specialist matches. Always surfaces the routing decision so users can override. Triggers — "research [topic]", "look into [topic]", "what do we know about [topic]", "investigate [topic]", "find me information on [topic]", "do some research on [topic]", "I need to understand [topic]", or any research request that doesn't obviously match a more-specific specialist skill. Output is a markdown briefing (default) or .docx document (on request) with full citations and an audit log.
pulse
Multi-source recency research skill that takes the pulse of any topic across Reddit, Hacker News, the open web, and optionally X/Twitter within a configurable recent window (default 30 days). Forcing intake clarifies topic specificity, angle (trend/sentiment/problems/opportunities/comparison), time window, and platform scope before searching. Returns a synthesized briefing with citations, engagement metrics, and cross-platform pattern analysis. Triggers: 'pulse on [topic]', 'what's happening with [topic]', 'what are people saying about [topic]', 'current conversation about [topic]', 'take the pulse of [topic]', 'trending: [topic]', 'find me info on [topic]', or any variation requesting multi-source recency intelligence on a topic. Also use for competitor research, trend discovery, tool comparisons, and audience sentiment analysis.
patent
Patent prior-art and landscape intelligence skill — not generic patent help. Commits to one of five sub-use-cases via forcing intake (novelty search / freedom-to-operate / competitive landscape / acquisition diligence / litigation prior-art) before any search runs. Searches Google Patents, Espacenet, USPTO, and optionally Lens.org for citation-graph signals. Output is an editable Word document (.docx) with verdict, ranked closest art (claim-text extracted), CPC-class-aware landscape, family-resolved hits, geographic coverage, FTO flags where applicable, strategy recommendations, and full audit log. Triggers: 'prior art search for [invention]', 'patent search on [topic]', 'freedom to operate analysis', 'FTO for [product]', 'patent landscape for [field]', 'is [invention] novel', 'patents on [topic]', 'competitive patent analysis', 'prior art for litigation', 'patent diligence on [company]'. Produces search signal, not legal advice — always recommends consulting a patent attorney before filing or licensing decisions. Trademark, copyright, and trade-secret questions are out of scope.
notebooklm
Browser automation skill for controlling Google's NotebookLM. Handles reading and querying notebooks, adding sources (URLs, text, files, YouTube links, synthesized content), generating Studio outputs (Audio Overview, infographics, slide decks, study guides, briefing docs, mind maps, timelines, FAQs), and creating new notebooks. Triggers on any phrase involving NotebookLM — 'open NotebookLM', 'check my [name] notebook', 'pull info from NotebookLM', 'ask my notebook about X', 'add [source] to NotebookLM', 'create an infographic in NotebookLM', 'use NotebookLM Studio', 'generate a slide deck from my notebook', or any variation where the goal involves NotebookLM. Requires browser automation environment — fails gracefully when unavailable.
litreview
Academic literature orientation skill that searches papers via Consensus, builds a strategic search plan using PICO (default) or SPIDER / Decomposition / hybrid as fallbacks, and synthesizes findings into a professionally formatted Word document (.docx) research guide. Grill-me intake (research question specificity + framework hint + tentative depth) before the recon search; a second forcing checkpoint after Phase 2 confirms framework + sub-areas + depth before searches consume budget. Configurable depth (5/10/20 queries) controls coverage vs. speed. Output is a 'launching pad' — not a finished review, but an orientation guide that lets a researcher dive in confidently. Triggers: 'litreview on [topic]', 'literature review on [topic]', 'I'm starting a literature review on X', 'I'm writing a paper on X', 'help me research X', 'I'm doing research on X', 'can you help me research X'. Do NOT trigger for single one-off paper searches where the user just wants a quick list — that's a plain Consensus search.