Social Ads

Social Ads Generator

Overview

Generate platform-ready image ads with multiple variations for Instagram, Facebook, LinkedIn, and more. Create and post directly via social-manager skill.

You generate the ads yourself using an AI image provider for images, Python Pillow for text overlays, and social-manager skill for posting.

Dependencies

This skill requires the following:

Dependency	Required?	Why
social-manager skill	Yes — for posting ads directly to social platforms
google-workspace skill	Optional — for saving ad assets to Google Drive
gemini-api skill	If user chooses Gemini — for image generation via Gemini Imagen
$GEMINI_KEY (stand-in, use with vault-curl)	If user chooses Gemini — for image generation via Gemini Imagen
$OPENAI_KEY (stand-in, use with vault-curl)	If user chooses OpenAI — for image generation via DALL-E 3
Pillow Python library	Yes — for adding text overlays, resizing, and image processing

Before starting, check:

1. Is social-manager skill installed and social accounts connected in the dashboard?
2. Which image keys do you have? [ -n "$OPENAI_KEY" ] && echo openai; [ -n "$GEMINI_KEY" ] && echo gemini. These are stand-ins (mcb_sec_…) for the user's keys, not the keys: they only work through vault-curl. A key added after you started is not in your env yet: run vault-curl --secrets and use an entry's surrogate string in place of $OPENAI_KEY / $GEMINI_KEY. Agents not yet moved to the vault may instead have real keys in the OPENAI_API_KEY / GEMINI_API_KEY variables: use those with plain curl, in the same header. Never print a key or stand-in, and never use OpenClaw's built-in secrets tool (these keys are not there). Missing keys: ask the user to add them in MyClawBots → Settings → Integrations; never ask them to paste a key into chat.
3. Is Pillow installed? Run pip install Pillow if needed.

Provider Selection — ASK THE USER

Before generating any ad images, you must ask the user which AI provider to use:

"I can generate ad images using:

1. OpenAI DALL-E 3 — high quality, ~$0.04/image, great for photorealistic product shots
2. Google Gemini Imagen — cheaper (~$0.025/image), supports image editing
3. Use existing image — upload your own and I'll add text overlays with Pillow

Which would you prefer?"

If only one key is available, inform the user and use that one. If neither key is set, default to offering existing image upload (free).

Capabilities

• Generate ad images via DALL-E 3 API
• Add text overlays with Python Pillow
• Create multiple variations for A/B testing
• Resize for platform specs (IG, FB, LinkedIn, Twitter)
• Post directly via social-manager skill
• Track performance basics

Step-by-Step Workflow

1. Gather Ad Brief

• Product/service name and description
• Target audience
• Key benefit/value proposition
• CTA (Call-to-Action): Shop Now, Learn More, Sign Up, Try Free
• Brand colors (primary, secondary)
• Budget and timeline

2. Generate Ad Image

Option A: OpenAI DALL-E 3 (if user chose OpenAI)

vault-curl -sS https://api.openai.com/v1/images/generations \
  -H "Authorization: Bearer $OPENAI_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "dall-e-3",
    "prompt": "Professional e-commerce ad for blue light blocking glasses. Product shot with lifestyle background. Modern, clean aesthetic. Person wearing glasses at computer. Bright, well-lit studio photography. Background: soft blurred office environment. Colors: blue (#1a73e8) and white accents. No text, professional product photo style.",
    "size": "1024x1024",
    "quality": "hd",
    "n": 1
  }' | jq -r '.data[0].url'

Save the image:

curl -s [IMAGE_URL] -o ad_image_1.png

Option B: Google Gemini Imagen (if user chose Gemini)

Easiest, and always works: the gemini-image skill (claw install gemini-image once). It finds the key itself and calls Gemini through the vault:

python3 /root/.openclaw/skills/gemini-image/scripts/gemini_image.py \
  --prompt "Professional e-commerce ad for blue light blocking glasses. Lifestyle product shot, no text." \
  --out /root/.openclaw/workspace/media/ad_image_1.png

To edit an existing image, add --input-image PATH. The Python SDK below sends the stand-in directly, so it works only when AGENTS.md says plain curl and SDKs work (the MyClawBots vault proxy is on).

With the vault proxy on, Python:

from google import genai
from google.genai import types
import os, base64

client = genai.Client(api_key=os.environ["GEMINI_KEY"])  # needs the vault proxy

response = client.models.generate_content(
    model="gemini-3-pro-image-preview",
    contents="Professional e-commerce ad for blue light blocking glasses. Product shot with lifestyle background. Modern, clean aesthetic. Person wearing glasses at computer. Bright, well-lit studio photography. Background: soft blurred office environment. Colors: blue (#1a73e8) and white accents. No text, professional product photo style.",
    config=types.GenerateContentConfig(
        response_modalities=["image", "text"],
    )
)

for part in response.candidates[0].content.parts:
    if part.inline_data:
        with open("ad_image_1.png", "wb") as f:
            f.write(base64.b64decode(part.inline_data.data))
        print("Saved ad_image_1.png")

3. Add Text Overlay with Python Pillow

Install Pillow:

pip install Pillow

Python script to add text:

from PIL import Image, ImageDraw, ImageFont
import textwrap

# Open base image
img = Image.open('ad_image_1.png')
width, height = img.size

# Create drawing context
draw = ImageDraw.Draw(img)

# Load font (use system font)
try:
    title_font = ImageFont.truetype("/usr/share/fonts/truetype/dejavu/DejaVuSans-Bold.ttf", 60)
    body_font = ImageFont.truetype("/usr/share/fonts/truetype/dejavu/DejaVuSans.ttf", 36)
    cta_font = ImageFont.truetype("/usr/share/fonts/truetype/dejavu/DejaVuSans-Bold.ttf", 40)
except:
    title_font = ImageFont.load_default()
    body_font = ImageFont.load_default()
    cta_font = ImageFont.load_default()

# Add semi-transparent overlay for text readability
overlay = Image.new('RGBA', img.size, (0, 0, 0, 150))
img = Image.alpha_composite(img.convert('RGBA'), overlay).convert('RGB')
draw = ImageDraw.Draw(img)

# Add headline
headline = "Protect Your Eyes\nBoost Your Focus"
text_bbox = draw.textbbox((0, 0), headline, font=title_font)
text_width = text_bbox[2] - text_bbox[0]
text_x = (width - text_width) // 2
draw.text((text_x, 80), headline, fill=(255, 255, 255), font=title_font)

# Add description
description = "Premium blue light blocking lenses"
text_bbox = draw.textbbox((0, 0), description, font=body_font)
text_width = text_bbox[2] - text_bbox[0]
text_x = (width - text_width) // 2
draw.text((text_x, 280), description, fill=(240, 240, 240), font=body_font)

# Add CTA button background
button_y = 450
button_height = 80
draw.rectangle([100, button_y, width-100, button_y+button_height], fill=(26, 115, 232))

# Add CTA text
cta_text = "SHOP NOW"
text_bbox = draw.textbbox((0, 0), cta_text, font=cta_font)
text_width = text_bbox[2] - text_bbox[0]
text_x = (width - text_width) // 2
draw.text((text_x, button_y+15), cta_text, fill=(255, 255, 255), font=cta_font)

# Save
img.save('ad_image_1_with_text.png', quality=95)
print('✓ Ad image created: ad_image_1_with_text.png')

Run it:

python add_ad_text.py

4. Create Variations

Generate 3-5 variations with different:

• Copy: Different headlines, benefits, angles
• Colors: Different primary colors
• Layout: Text position, button style
• Style: Minimalist, bold, lifestyle, product-focused

For each variation, repeat step 2-3 with modified DALL-E prompt or Pillow overlay.

5. Resize for Each Platform

Platform specs:

• Instagram Feed: 1080x1350 (4:5)
• Instagram Story: 1080x1920 (9:16)
• Facebook Feed: 1200x628 (1.91:1)
• LinkedIn: 1200x627 (1.91:1)
• Twitter: 1600x900 (16:9)
• TikTok: 1080x1920 (9:16)

Use ImageMagick or PIL to resize:

from PIL import Image

def resize_for_platform(input_file, platform):
    img = Image.open(input_file)

    specs = {
        'instagram_feed': (1080, 1350),
        'instagram_story': (1080, 1920),
        'facebook': (1200, 628),
        'linkedin': (1200, 627),
        'twitter': (1600, 900),
    }

    width, height = specs[platform]
    img_resized = img.resize((width, height), Image.Resampling.LANCZOS)
    img_resized.save(f'ad_{platform}.png', quality=95)
    print(f'✓ {platform}: ad_{platform}.png')

# Resize for all platforms
for platform in ['instagram_feed', 'instagram_story', 'facebook', 'linkedin', 'twitter']:
    resize_for_platform('ad_image_1_with_text.png', platform)

6. Post to Social Media

Use social-manager skill (already installed):

# Instagram
social-manager +post \
  --platform instagram \
  --image ad_instagram_feed.png \
  --caption "Protect your eyes, boost your focus 👀 Premium blue light blocking glasses. Shop now with code SAVE10 #BlueLight #EyeHealth"

# Facebook
social-manager +post \
  --platform facebook \
  --image ad_facebook.png \
  --caption "Introducing our premium blue light glasses. Reduce eye strain by 85%. Shop now →"

# LinkedIn
social-manager +post \
  --platform linkedin \
  --image ad_linkedin.png \
  --caption "How I protect my eyes during long work sessions: premium blue light blocking glasses. If you spend 8+ hours at a screen, these are worth trying. Link in bio."

# Twitter
social-manager +post \
  --platform twitter \
  --image ad_twitter.png \
  --caption "Spending all day in front of screens? Protect your eyes with our premium blue light glasses. 85% less eye strain. Shop now 🛒"

7. Monitor Performance

Track metrics manually or via platform analytics:

• Click-through rate (CTR): % of people who clicked
• Conversion rate: % of clicks that became sales
• Cost per result: Total spend / conversions
• Engagement: Likes, comments, shares

Platform Specifications & Best Practices

Instagram Feed (1080x1350)

• Best times: 11am, 7-9pm EST
• Max caption: 2,200 chars
• Hashtags: 5-30 relevant tags
• Copy style: Aspirational, lifestyle-focused

Facebook (1200x628)

• Best times: 1-3pm EST, 7-9pm EST
• Max caption: 125 chars (visible before "more")
• Copy style: Benefit-driven, pain-point focused
• CTA button options: Learn More, Shop Now, Sign Up

LinkedIn (1200x627)

• Best times: 8-10am EST, Tuesday-Thursday
• Max caption: 3,000 chars
• Copy style: Professional, B2B, thought leadership
• Avoid: Aggressive sales language

Twitter (1600x900)

• Best times: 9am, 5pm EST
• Max text: 280 chars (including link)
• Copy style: Conversational, quick value prop
• Engagement: Reply to comments immediately

Cost

• DALL-E 3 image generation: ~$0.04-0.08 per 1024x1024 image (HD)
• Pillow image processing: Free (open source)
• Social media posting: Free (via social-manager skill)

Total cost per campaign: $0.20-0.50 for 3-5 concept images + variations.

Pro Tips

• A/B test variations: Always test 3-5 versions to find what resonates
• Use contrasting colors: Make CTA button pop against ad background
• Clear value prop: Lead with the biggest benefit, not features
• Mobile-first: Test how ads look on phone (narrow screens)
• People + product: Lifestyle ads (people using product) outperform product-only shots
• Test different CTAs: "Shop Now" vs "Learn More" vs "Try Free" can have big impact
• Posting schedule: Post when audience is most active (platform-specific)
• Negative space: Don't overcrowd the ad; whitespace draws attention to key elements
• Font legibility: Test on small screens; use bold, clean fonts
• Compliance: Follow platform ad policies (no misleading claims, clear disclaimers)