By Mike Demo Demopoulos, GBTI Network Supporter. Originally published on gbti.network.
Magic Manta is a social radar scanner. Enter a company's domain and it finds that company's social media profiles across X, Threads, Instagram, LinkedIn, Bluesky, GitHub and more, then shows the evidence behind every handle it returns.
How a lookup works
It reads the site's Schema.org sameAs data and the links on its official pages, then falls back to targeted search for anything those do not declare. A profile declared through JSON-LD sameAs or an official-site DOM link gets a Verified badge, with its evidence source shown. A search fallback candidate stays Unverified unless it links back to the official domain. Every result carries a last-checked timestamp, and conflicts between official sources are flagged.
Beyond one domain
Batch lookups sweep several domains in a single run: five per batch on the Operative plan, twenty-five on Deep Recon.
For agents
The radar core is framework-agnostic, with three doorways into the same engine. The MCP server is live today, auth-gated, at 100 lookups per account per day. A REST endpoint and a React SDK are listed as coming soon, and there is no public REST API yet.
Accounts and plans
Single lookups need no account. An account unlocks saved lookups, lists, paid plans and the MCP server, with sign-in through Google, Apple, Microsoft, AgentID, or email and password. Single lookups are free. Operative is a one-time five dollars for an account, lookup history, saved lists and CSV or JSON export. Deep Recon is ten dollars a month, and Brand Command twenty dollars a month.
Hosting and carbon
The site runs on Lovable Cloud behind Cloudflare, which the Green Web Foundation verifies as green hosting, and 1.5 percent of purchases goes to Stripe Climate for carbon removal. A machine-readable disclosure sits at /carbon.txt.
The resource shared in this post is a new project published by GBTI Network member Mike Demo Demopoulos.
Are you a writer, musician, or product developer? We would love to support your work on the GBTI Network. For more information about how to join our community visit https://gbti.network
To follow Mike Demo Demopoulos's work more closely, consider joining our network and subscribing to them directly: https://gbti.network/members/mike-demo/
Top comments (1)
Verification is the hard half of this problem — finding candidate profiles is string search, deciding which ones are the company is evidence work.
From running public-channel monitoring at scale, the cheap heuristics that caught the most fakes for me: engagement-to-subscriber ratio (a "verified" channel with 10k subscribers and 40 views per post is a bought audience), first-post archaeology (real brands have boring 2016 posts; clones start the week they were needed), and cross-referencing the handle spelling across platforms — registrars eat typos, and the gap between
acme-ioandacmeiois where the squatters live.The one class none of these catch: an insider moving house before a launch. No public signal flags that one — which is why I now treat "consistent since the beginning" as the strongest signal I can actually verify, and say so in the report.