Image Trust Protocol

Multi-source fact-check verification for AI agents. Synthesized verdicts from credentialed publishers, with full attribution.

For AI agent builders

Add image and claim verification to your AI agent in one paste. Our agent setup instructions teach your agent when to call us and how to surface results with proper citations.

View agent setup instructions →

What this is

Image Trust Protocol is a verification API that returns multi-source fact-check verdicts on images and claims. We synthesize verdicts from PolitiFact, FactCheck.org, dpa, PesaCheck, and a growing set of credentialed fact-checkers, with full attribution preserved on every response.

Three lookup endpoints, one consistent response shape:

GET /v1/lookup/hash/{hash} — by 64-bit perceptual hash
POST /v1/lookup/image — by raw image bytes
GET /v1/lookup/claim?text=... — by claim text, full-text search across 50+ languages

Distribution surfaces

REST API

Standard B2B API. Authenticate via your operator key. https://api.imagetrustprotocol.world/v1

MCP server

Model Context Protocol server exposing the same three lookups as MCP tools. For agent platforms that consume MCP natively (Claude Code, Cursor, Codex, Amp, Anthropic directory).

Agent setup instructions

Natural-language prompt designed for direct paste into your agent's system prompt or skill. Get the instructions.

llms.txt

AI-readable site summary at /llms.txt following the emerging convention.

What we don't do

We don't generate fact-checks. We surface verdicts already published by credentialed organizations.
We don't detect AI-generated images via pixel analysis. We surface what fact-checkers have concluded about specific images.
We don't store user images. Only perceptual hashes are retained.
We don't speak for the publishers we surface. Their verdicts are theirs; their original articles linked from every response.