ARGUS: authenticity scoring for social profiles and posts
ARGUS gives a social media profile or post a 0 to 100 authenticity score from image, text, behavior and network signals. Each score is an algorithmic opinion, a person reviews any report before it goes public, and anyone can dispute one.
- Tool
- ARGUS
- Built
- Mar 2026
- Code
- Public on GitHub
- Stack
- JavaScript, Cloudflare Pages, Workers, KV and R2, Anthropic Claude
- Reports
- Algorithmic opinion, human-reviewed, disputable
The idea
Fake accounts, bot engagement and AI-generated profiles distort ad results and lead validation. Advertisers need a quick way to judge whether an account or a burst of comments looks real.
Judging real people's profiles is sensitive. Any tool that does it has to be clear that it gives an opinion, not a verdict, and let people respond.
How we built it, step by step
- Built 4 detection enginesARGUS combines network (35%), behavioral (30%), image (20%) and text (15%) signals into a deterministic weighted score, minus listed penalties.
- Covered the main networksIt supports LinkedIn, Reddit, Instagram, X, Facebook, YouTube, Quora, TikTok and Pinterest. A private-profile mode works from pasted content and screenshots.
- Added content and commenter analysisClaude reviews the content, and a commenter-level check looks for bot activity and coordinated behavior.
- Put a person in the loopNo public report goes live until a person approves it. Every report is framed as an algorithmic opinion.
- Built a dispute processA dispute and opt-out page lets anyone challenge a report and upload evidence.
- Shipped on CloudflareARGUS runs on Cloudflare Pages Functions, KV, R2 and Workers at a stated cost of about $0 to $5 a month. It includes a Chrome extension, public report pages with a sitemap and an admin dashboard.
What it does
- 0 to 100 authenticity score
- Network, behavioral, image and text signals
- Support for 9 social networks
- Private-profile mode from pasted content
- Human approval before any public report
- Dispute and opt-out page
- Chrome extension
- Results by email
How people find it
- Free hosted tool on ahmeego.com
- Public source on GitHub
- Chrome MV3 extension
- Public report pages with a dynamic sitemap
Questions people ask about this work
How can I tell if a LinkedIn profile is fake?
Look at the network around it, how it behaves over time, its images and its writing together, since one signal alone is weak. ARGUS weighs those 4 signals into a 0 to 100 score. Treat the result as an opinion to check, not proof.
How do I check if comments on my ads are from bots?
Look for patterns across commenters, such as similar timing, wording or account history. ARGUS runs a commenter-level check for bot activity and coordination on a post.
What can I do if ARGUS has a report about my profile?
Use the dispute and opt-out page to challenge it and upload evidence. Reports are algorithmic opinions, and a person reviews each one before it is published.
Services in this project
Related projects
Business Discovery Engine
The engine finds local businesses in public directories, visits their websites for contact details and social links, and writes each batch to Google Sheets as it runs.
Marketing Analytics Auditor
The auditor scans a page for 10 marketing tags and consent setup, audits a GA4 property through the Admin API and writes the code to fix what it finds.
Creative Asset Validator and Ad Creative MCP
Upload one image and the validator checks it against specs for 50+ advertising platforms, then resizes it into each size you are missing.
Built and written up by John M. Williams, founder of It All Started With An Idea in Queen Creek, AZ. 15+ years and $350M+ in managed ad spend.