Glasses to Social

Turn smart glasses photos into social media posts. Monitors a Google Drive folder for new images from Meta Ray-Ban glasses (or any smart glasses), analyzes them with vision AI, drafts tweets/posts in the user's voice, and publishes on approval. Use when setting up a glasses-to-social pipeline, processing smart glasses photos for social media, or creating hands-free content workflows.

Security Scan

The skill's files and instructions are mostly coherent with its stated purpose, but there are privacy- and scope-related concerns (notably the recommendation to make a Google Drive folder publicly viewable and vague instructions for vision analysis/posting) that the user should understand before installing.

Purpose & Capability

Name/description match the actual artifacts: scripts and SKILL.md implement a pipeline that downloads photos from a Google Drive folder, marks processed files, asks a vision model to analyze images, and drafts posts for social media. The required tools mentioned (gdown, a vision model, Twitter/X credentials) are appropriate for the declared goal.

Instruction Scope

SKILL.md instructs the user to share the Drive folder with 'Anyone with link can view' (privacy exposure) and uses gdown to pull files without describing authenticated access or access controls. The instructions for 'analyze with vision' and 'publish' are vague and leave the agent broad discretion (e.g., which model/service to call, where drafts are stored/transmitted). The included scripts only perform downloads, file enumeration, and update a processed list (no hidden data exfiltration), but the recommendation to make the folder public materially broadens exposure risk.

Install Mechanism

No install spec is provided (instruction-only), which minimizes install risk. The scripts rely on standard CLI tools (gdown, jq, find, cp) but the package installation steps are not included. There are no downloads from arbitrary URLs or extract actions.

Credentials

The skill requests no environment variables and doesn't demand credentials in its files. It does note that Twitter/X credentials are required for posting and that a vision-capable model/service is needed; both are proportional. However, SKILL.md's approach (making Drive folder public rather than using authenticated Drive API access) is a questionable design choice that trades credential management for public exposure.

Persistence & Privilege

The skill is not marked always:true and does not request persistent system privileges. The scripts operate only within the configured download and data directories and modify their own processed.json file; they do not change other skills or system-wide configurations.

What to consider before installing

This skill is generally coherent with its stated purpose, but pause before installing and consider privacy and access controls: - The SKILL.md tells you to set the Google Drive folder to 'Anyone with link can view'. That makes your photos publicly accessible to anyone who gets the link. Prefer a private folder and use authenticated Drive access (OAuth or a service account) instead of a public share. - The scripts require gdown and jq but the skill doesn't provide installation steps — ensure you install these from trusted package sources and run the scripts in a controlled directory. - The instructions for image analysis and posting are vague: decide which vision service/model will run and where drafts are stored/transmitted. If you use a third-party vision API, review its privacy policy and avoid sending sensitive photos. - Posting automation requires social media credentials. Use a dedicated account or app token with limited scope and rotate/revoke keys if needed. Ensure the agent asks for explicit approval before posting (autoPost=false by default is safer). - Consider removing/examining EXIF/metadata from photos before uploads and obtain consent from people who appear in images. - If you want tighter security, modify the pipeline to use Google Drive API with OAuth/service account rather than gdown + public shares, and add explicit logging and review steps. Confidence in this assessment is medium. If you want a higher-confidence verdict, provide answers or changes to: how Google Drive access will be authenticated (public link vs OAuth/service-account), which vision service will be used, and the exact posting mechanism for social platforms.

Transform photos from smart glasses into social media posts with AI-generated captions.

Overview

This skill creates a pipeline from smart glasses (Meta Ray-Ban, etc.) to social media:

1. User snaps photo with glasses
2. Photo syncs to Google Drive folder
3. Agent detects new photo, analyzes with vision
4. Agent drafts post matching user's voice/style
5. User approves, agent publishes

Setup

1. Configure Google Drive Folder

Create a shared Google Drive folder for glasses photos:

# User creates folder "Glasses-to-Social" in Google Drive
# Share with "Anyone with link can view"
# Copy the folder URL

2. Set Up Config

Create config file at glasses-to-social/config.json:

{
  "googleDriveFolderUrl": "https://drive.google.com/drive/folders/YOUR_FOLDER_ID",
  "folderId": "YOUR_FOLDER_ID",
  "downloadPath": "./glasses-to-social/downloads",
  "processedFile": "./glasses-to-social/data/processed.json",
  "defaultHashtags": ["#MedicalAI", "#HealthTech"],
  "autoPost": false
}

3. Configure Glasses Auto-Sync

For Meta Ray-Ban glasses:

1. Open Meta View app on phone
2. Settings > Gallery > Enable "Import Automatically"
3. iOS: Enable Google Photos backup (syncs Camera Roll)
4. Create iOS Shortcut to copy new Meta photos to Google Drive folder

Usage

Manual Check

Ask the agent to check for new photos:

Check my glasses folder for new photos

Automated Monitoring

Set up a cron job to check periodically:

{
  "name": "Glasses-to-Social: Check photos",
  "schedule": {"kind": "cron", "expr": "*/15 * * * *", "tz": "UTC"},
  "payload": {
    "message": "Check the Glasses-to-Social folder for new photos. If found, analyze and draft a tweet."
  }
}

Processing Flow

When a new photo is detected:

1. Download from Google Drive using gdown:

   gdown --folder "FOLDER_URL" -O ./downloads/ --remaining-ok
   

2. Compare against processed list in data/processed.json

3. For new photos, analyze with vision:

   • Describe the scene/subject
   • Identify relevant context for social post
   • Note any text, people, or notable elements

4. Draft post matching user's voice:

   • Keep it concise and authentic
   • Add relevant hashtags
   • First-person perspective works well for glasses content

5. Send draft to user for approval:

   • Include image preview
   • Show proposed caption
   • Wait for "POST" confirmation or edits

6. On approval, publish to configured platform (X/Twitter, etc.)

7. Mark photo as processed in data/processed.json

Scripts

check-new-photos.sh

Checks Google Drive folder for new images:

scripts/check-new-photos.sh

Output format when new photo found:

NEW_PHOTO_PATH:/path/to/downloaded/photo.jpg

File Tracking

Track processed photos in data/processed.json:

{
  "processed": ["photo1.jpg", "photo2.jpg"],
  "pending": []
}

Tips

• First-person POV content performs well ("Look what I just saw...")
• Keep captions authentic, not overly polished
• Works great for conferences, interesting sightings, daily moments
• Consider time-of-day context when drafting

Requirements

• gdown Python package for Google Drive access
• Vision-capable model for image analysis
• Twitter/X credentials for posting (optional)