Insta Content Engine

Find trending topics, create editorial-style social media graphics, and post to X/Twitter and Instagram. Includes image generation with photographic backgrou...

Security Scan

The skill mostly does what its name says, but there are mismatches and operational practices that could expose credentials and access unrelated local config; review before using with real accounts.

Purpose & Capability

The files and SKILL.md implement the stated functionality (viral search, image generation via OpenAI, and Instagram posting). However the registry metadata declares no required env vars or credentials while the SKILL.md and scripts clearly require OPENAI_API_KEY, BRAVE_API_KEY, and IG_USERNAME/IG_PASSWORD — this mismatch is unexpected and incoherent with the published metadata.

Instruction Scope

Runtime instructions and code are explicit about network calls to OpenAI and Brave Search and about using the bird CLI and instagrapi. Concerns: (1) instagram-search.js will try to read ~/.openclaw/clawdbot.json to extract an API key (accesses a potentially broad local config), (2) instagram-post.js constructs a Python script embedding username/password and executes it via python -c which can expose credentials on the process command line, and (3) scripts create and read a session file at ~/.openclaw/ig_session.json (persistent tokens). These behaviors go beyond simple 'search and post' and have privacy/credential exposure implications.

Install Mechanism

No install spec and all code is shipped in the skill bundle (instruction-only install). No external archives or unknown download URLs are used. Required runtimes (node, python, pip packages, bird CLI) are standard for the task.

Credentials

The credentials requested by the code (OpenAI key for image generation, Brave search key for Instagram scraping, and Instagram username/password for posting) are proportionate to the functionality. However the skill registry advertises no required env vars, so the runtime expectation is not declared to users. Also reading ~/.openclaw/clawdbot.json can surface unrelated credentials from a user's OpenClaw config.

Persistence & Privilege

The skill stores Instagram session data under ~/.openclaw/ig_session.json so the login can be reused. This is expected for a posting tool, but it creates persistent tokens on disk that should be protected. The skill is not force-installed (always: false) and does not modify other skills, but it does access a shared OpenClaw config path which could overlap with other tooling.

What to consider before installing

This skill implements searching, image generation (OpenAI), and Instagram posting, but before installing you should: (1) expect to provide OPENAI_API_KEY, BRAVE_API_KEY, and IG_USERNAME/IG_PASSWORD — the published metadata does not declare these, so be cautious, (2) review and, if needed, sanitize ~/.openclaw/clawdbot.json because the script reads it for keys, (3) avoid running with your main Instagram account credentials — consider a throwaway/test account, (4) be aware instagram-post.js embeds credentials into a Python command string which can expose them in process listings while running, (5) note the skill will persist session tokens to ~/.openclaw/ig_session.json — protect or remove that file if you stop using the skill, and (6) inspect the code yourself or run it in an isolated environment (container or VM) if you plan to use real credentials. If you want to proceed safely, ask the author to update the registry metadata to list required env vars and to change the posting flow to avoid embedding credentials on the command line (e.g., pass via stdin or use a secure session flow).

Social Search & Content Creator

Find → Create → Post pipeline for X/Twitter and Instagram content.

Prerequisites

• X/Twitter: bird CLI installed and authenticated via browser cookies
• Instagram Search: Brave Search API key (set BRAVE_API_KEY or configure in OpenClaw)
• Image Generation: OpenAI API key (set OPENAI_API_KEY) — uses gpt-image-1
• Instagram Posting: pip3 install instagrapi pillow + IG_USERNAME/IG_PASSWORD env vars

1. FIND — Research Trending Topics

X/Twitter (bird CLI)

bird search "query" -n 10          # Search tweets
bird read <url-or-id>              # Read a specific tweet
bird thread <url-or-id>            # Full thread
bird mentions                      # Check mentions
bird user-tweets <username>        # User's tweets

Instagram (Brave-powered scraper)

node {baseDir}/scripts/instagram-search.js "query" [--limit 10] [--type posts|profiles|reels|all]

Types: posts (inurl:/p/), profiles (user pages), reels (inurl:/reel/), all (default)

Viral/Trending Search (recommended starting point)

# Find viral content about a topic (sorted by engagement)
node {baseDir}/scripts/viral-search.js "query" [options]

# Options:
#   --platform x|instagram|both    (default: both)
#   --min-likes N                  Min likes to include (default: 50)
#   --min-retweets N               Min retweets (default: 10)
#   --sort engagement|recent       (default: engagement)
#   --limit N                      Results per platform (default: 10)
#   --trending                     What's blowing up RIGHT NOW on X
#   --days N                       Only last N days (default: 7)

Examples:

# What's viral about AI trading this week
node {baseDir}/scripts/viral-search.js "AI trading" --min-likes 100 --days 7

# What's trending across crypto right now (no query needed)
node {baseDir}/scripts/viral-search.js --trending

# Instagram-only viral search
node {baseDir}/scripts/viral-search.js "DeFi" --platform instagram --limit 10

# Lower the bar for niche topics
node {baseDir}/scripts/viral-search.js "delta neutral farming" --min-likes 10 --min-retweets 3

How engagement scoring works:

• Score = likes × 1 + retweets × 3 + replies × 2
• Retweets weighted highest (distribution signal)
• Replies weighted over likes (conversation signal)
• Results sorted by score descending

2. CREATE — Generate Editorial Graphics

The System: Photographic Background + Dark Gradient + Bold Typography

This is what makes the images look professional, not just text on a background.

The approach:

• Top 60%: Cinematic photographic scene (real subjects, dramatic lighting)
• Bottom 40%: Heavy dark gradient overlay (95% black → transparent)
• Text: Bold condensed typography (Anton-style) sitting on the gradient
• Result: Looks like Bloomberg/Kalshi editorial cards

Generate Images

OPENAI_API_KEY="your-key" python3 {baseDir}/scripts/gen_image.py "prompt" output.png [1024x1536]

Sizes: 1024x1024 (square), 1024x1536 (portrait/Instagram), 1536x1024 (landscape/Twitter)

⚠️ CRITICAL IMAGE RULES (learned the hard way)

1. NEVER just put text on a solid dark background — it looks flat and amateur
2. ALWAYS use a photographic/cinematic background — real scenes, real subjects
3. ALWAYS use dark gradient overlay on the lower 40% for text readability
4. Keep subject visible in the top 60% — face, setting, context
5. Text lives in the bottom 40% only — never cover the subject
6. Use bold condensed typography (Anton-style) — not regular weight fonts
7. Include branded elements — badge, branding URL, slide counter
8. Cinematic lighting — moody, dramatic, not flat studio lighting

Master Prompt Template

Copy this template exactly. Only change the parts in [BRACKETS].

Instagram carousel post design. EXACTLY 1080 pixels wide by 1350 pixels tall (4:5 vertical ratio).

Background: [DESCRIBE THE SCENE - who/what is shown, setting, lighting, mood]. Keep subject clearly visible in top 60% of image. Cinematic lighting, dramatic.

Overlay: Heavy dark gradient ONLY on bottom 40% of image - solid black (95% opacity) at very bottom, fading to transparent at 60% height. Top portion stays clear.

Text positioning (bottom area):
- 65% from top: Small rectangular badge with [COLOR] background and [TEXT COLOR] bold text: [BADGE TEXT]
- 70-90% from top: Huge white bold text in Anton font, uppercase, centered:
[MAIN HEADLINE
SPLIT ACROSS
2-3 LINES]
- [Optional: Below headline in smaller white text, any supporting details]
- Bottom right corner (95% from top): [X/Y] in white, small
- Bottom left corner (95% from top): [brand.url] in white, Montserrat Bold, small, lowercase

Font: Anton for headline. Montserrat Bold for branding.

Style: Clean Instagram carousel slide, vertical format, text at bottom like Kalshi/Bloomberg editorial style. Cinematic, high contrast, professional.

Badge Color System

• 🔵 Cyan (#00D4FF): "EXPLAINED", "GUIDE", educational content
• 🟠 Orange (#FF6B00): "HOW IT WORKS", "DEEP DIVE", breakdowns
• 🟢 Green (#22c55e): "THE SHIFT", "SOLUTION", "REAL CASE", positive takes
• 🔴 Red (#e63946): "WARNING", "UNFAIR", "MYTH", negative/urgent
• ⚫ Black (#000000): "TRENDING", "BREAKING", news

Background Scene Ideas by Topic

Crypto/Trading: Trader at multiple monitors, trading floor, candlestick charts on screens Tech/AI: Neural network visualizations, server rooms, holographic displays Finance: City skyline at night, office with Bloomberg terminals, stock exchange Startup: Founder at whiteboard, team in modern office, product on screen Education: Person studying, lecture hall, books with highlighted text

Carousel Structure (3 slides)

• Slide 1 — The Hook: Sets the scene, asks the question (e.g., "What is X?")
• Slide 2 — The Breakdown: How it works, the layers, the details
• Slide 3 — The Shift: Why it matters, the bigger picture, call to action

Example: Crypto Educational Carousel

Slide 1 prompt background: "A young male trader in a dark room lit by multiple monitors showing crypto charts. Blue and green screen glow on his face. Hoodie, leaning forward, focused."

Slide 2 prompt background: "Close-up of an AI neural network visualization on a large curved monitor in a dark trading room. Neon blue data streams flowing across the screen."

Slide 3 prompt background: "Aerial view of a massive trading floor with hundreds of screens running autonomously, seats empty. No humans present. Dramatic overhead shot, neon blue glow."

3. POST — Publish Content

X/Twitter (bird CLI)

bird tweet "Your tweet text"                         # Text only
bird tweet "Text" --media image.jpg                  # With image
bird tweet "Text" --media img1.jpg --media img2.jpg  # Multiple
bird reply <tweet-id-or-url> "Reply text"            # Reply

Instagram (instagrapi)

# Photo post
node {baseDir}/scripts/instagram-post.js --action photo --caption "Caption" image.jpg

# Reel
node {baseDir}/scripts/instagram-post.js --action reel --caption "Caption" video.mp4

# Story
node {baseDir}/scripts/instagram-post.js --action story --caption "Caption" image.jpg

# Carousel
node {baseDir}/scripts/instagram-post.js --action carousel --caption "Caption" img1.jpg img2.jpg img3.jpg

Instagram auth: Set IG_USERNAME and IG_PASSWORD env vars. Google sign-in won't work — need a password set on the account.

Full Workflow Example

# 1. Research what's trending
bird search "agentic trading crypto" -n 10
node {baseDir}/scripts/instagram-search.js "AI crypto trading" --limit 10

# 2. Generate 3-slide carousel
OPENAI_API_KEY="key" python3 {baseDir}/scripts/gen_image.py "slide 1 prompt..." slide1.png 1024x1536
OPENAI_API_KEY="key" python3 {baseDir}/scripts/gen_image.py "slide 2 prompt..." slide2.png 1024x1536
OPENAI_API_KEY="key" python3 {baseDir}/scripts/gen_image.py "slide 3 prompt..." slide3.png 1024x1536

# 3. Post to X
bird tweet "your caption" --media slide1.png --media slide2.png --media slide3.png

# 4. Post to Instagram
node {baseDir}/scripts/instagram-post.js --action carousel --caption "your caption" slide1.png slide2.png slide3.png