Paid Ads Generator
SkillMediaUse when you want paid ad copy for Google, Meta, or LinkedIn, including headlines, descriptions, and image briefs.
Available today. Use it from your connected AI after setup.
No other account needed.
Connect ahel once, and every AI you use reads what you have installed.
Then ask your AI: use the Paid Ads Generator skill
What this skill tells your AI
The instructions your AI receives, as published by j1ngg/tech-marketing-framework in .agents/skills/ads/SKILL.md and read by ahel’s review.
This skill transforms your product context into paid ad copy. It uses a creative matrix (Topics x Personas x Angles) to generate targeted variants for each platform.
Step 1: Platform Selection
Before generating any content, ask the user:
Platform:
- Google Ads (Search or Display)
- Meta (Facebook/Instagram)
- Multiple platforms
Ad format:
- Search (text only)
- Display/Feed (image + copy)
Wait for the user to answer before proceeding.
Step 2: Campaign Context
Ask the user:
- Campaign objective: Awareness, consideration, or conversion?
- Landing page URL: Where does the ad drive traffic?
- Specific offer or CTA: (Optional) Any promotion, demo offer, or specific action?
Wait for answers before proceeding.
Step 3: Gather Inputs
Read from repo automatically:
| File | What to extract |
|---|---|
docs/inputs/product_brief.md | Product name, 3 to 5 key capabilities, available links |
docs/inputs/messaging_positioning.md | Pain points (become Topics), value propositions |
docs/inputs/target_personas.md | Audience segments (become Personas) |
docs/inputs/competitor_intel.md | Competitor weaknesses, differentiation points |
Read these files before asking for additional inputs.
Then ask:
"Do you have recent competitive research to incorporate? If you ran how-they-market on a competitor, provide the file path (e.g.,
output/research/competitor-name.json). Otherwise, type 'skip' and I'll use the static competitor intel."
If the user provides a path, read that file and extract:
- Competitor messaging patterns
- Gaps in their positioning
- Angles they are not using
If the user types "skip" or the file does not exist, proceed with docs/inputs/competitor_intel.md only.
Step 4: Build the Creative Matrix
Construct the matrix from gathered inputs:
Topics (from messaging_positioning.md):
- Extract 3 to 4 pain points or value propositions
- Each becomes a core message theme
Personas (from target_personas.md):
- Extract 2 to 3 primary audience segments
- Each gets tailored language and emphasis
Angles (standard set):
- Problem-agitation: Lead with the pain, twist the knife
- Social proof: Lead with who else uses it, results they got
- How-to: Lead with the action, product enables it
- Comparison: Lead with competitor weakness, your strength (use competitor intel)
Matrix output: Topics x Personas x Angles
Do NOT generate all combinations. Select the 3 to 5 strongest combinations based on:
- Campaign objective (conversion = problem-agitation, social proof)
- Platform (LinkedIn = social proof works well for budget holders)
- Available proof points (only use social proof if testimonials exist)
Step 5: Generate Ad Copy
Generate 3 to 5 variants for each selected platform. Follow platform-specific rules below.
Platform Rules
Google Search Ads (Responsive Search Ads)
| Element | Spec | Guidance |
|---|---|---|
| Headlines | Up to 15, 30 chars each | Write 6 to 8. Front-load keywords. No punctuation at end. |
| Descriptions | Up to 4, 90 chars each | Write 3 to 4. Include CTA. One with social proof if available. |
| Display URL paths | 2 paths, 15 chars each | Use product name and action (e.g., /demo, /start) |
Google-specific rules:
- No exclamation marks in headlines
- No ALL CAPS
- No dynamic keyword insertion placeholders unless user requests
- Headlines should work independently (they get shuffled)
Meta Ads (Facebook/Instagram)
| Element | Spec | Guidance |
|---|---|---|
| Primary text | 125 chars recommended, 500 max | Hook in first line. Question or bold claim. |
| Headline | 27 chars recommended, 255 max | Benefit or CTA. Short wins. |
| Description | 27 chars recommended | Often hidden. Keep optional. |
| Image guidance | 1080x1080 or 1200x628 | Describe visual concept, not just "product screenshot" |
Meta-specific rules:
- No text covering more than 20% of image
- Primary text: First 125 chars must hook (rest gets truncated)
- Mobile-first: Assume thumb-stopping in feed
LinkedIn Ads
| Element | Spec | Guidance |
|---|---|---|
| Intro text | 150 chars before truncation, 600 max | Lead with insight or question. Business outcome focus. |
| Headline | 70 chars recommended, 200 max | Role-specific language. Speak to the budget holder. |
| Description | 100 chars recommended | Reinforce CTA. |
| Image guidance | 1200x627 | Professional. Data visualizations work well. |
LinkedIn-specific rules:
- Audience is VPs, Heads of, CTOs (budget holders, not builders)
- Focus on business outcomes: risk, cost, time, compliance
- Social proof from recognizable companies matters more here
- No hashtags in ad copy
Angle-Specific Templates
Problem-Agitation
[Pain point as question or statement]
[Consequence of not solving]
[Product as relief] + [CTA]
Social Proof
[Company/role] + [achieved outcome]
[How they did it with product]
[CTA: See how / Get the same]
How-To
[Action verb] + [desired outcome]
[Product makes it possible]
[CTA: Start / Try / See how]
Comparison
[Competitor approach] = [problem with it]
[Your approach] = [better outcome]
[CTA: Switch / Compare / See the difference]
Output Format
For each variant, output:
## Variant [N]: [Topic] x [Persona] x [Angle]
**Platform:** [Google/Meta/LinkedIn]
**Objective fit:** [Why this variant fits the campaign objective]
---
### Headlines
1. [Headline] (X chars)
2. [Headline] (X chars)
3. [Headline] (X chars)
[Continue as needed per platform spec]
### Descriptions
1. [Description] (X chars)
2. [Description] (X chars)
[Continue as needed per platform spec]
### Image/Creative Brief
[2 to 3 sentence description of visual concept. Include: subject, composition, mood, text overlay if any]
### Display URL / Paths
[For Google only]
---
After all variants, include:
## Summary
| Variant | Platform | Topic | Persona | Angle |
|---------|----------|-------|---------|-------|
| 1 | [Platform] | [Topic] | [Persona] | [Angle] |
| 2 | ... | ... | ... | ... |
**Recommended test priority:**
1. [Variant N] — [Why test this first]
2. [Variant N] — [Why test this second]
Review Checklist
Before delivering, verify each variant passes:
- Headlines under character limit for platform
- Descriptions under character limit for platform
- No dashes in copy (rewrite if needed)
- No passive voice
- No banned words (robust, seamless, leverage, cutting-edge, game-changer, revolutionary, transformative)
- No hedging ("we believe", "consider")
- Claims match product_brief.md exactly
- CTA matches campaign objective
- Social proof uses only approved customers from testimonials.md
- Comparison angle uses real competitor intel, not fabricated weaknesses
- Image briefs are specific enough to execute
Adaptation Notes
When docs are incomplete:
- If
product_brief.mdis empty, ask user for product name and 3 capabilities - If
messaging_positioning.mdis empty, ask user for 3 pain points - If
target_personas.mdis empty, ask user to describe 2 audience segments - If
competitor_intel.mdis empty AND user skips research file, omit comparison angle entirely
When testimonials are missing:
- Do not use social proof angle with specific company names
- Fall back to generic framing: "[Industry] teams" or "Engineering leaders"
When user wants more variants:
- Maximum 5 per platform per session
- If they want more, suggest running the skill again with different angle priorities
When user wants fewer variants:
- Minimum 2 to enable A/B testing
- Ask which angles to prioritize given their objective
Multi-platform requests:
- Generate variants for each platform sequentially
- Adapt the same Topic x Persona x Angle to each platform's specs
- Note where messaging must differ (LinkedIn = business outcome, Meta = hook fast)
Signals
- GitHub stars
- 56
- Forks
- 8
- Last commit
- Apr 2026
Advanced
- Catalog kind
- skill
- Gateway key
ads-j1ngg- Source
- github.com/j1ngg/tech-marketing-framework