Scry

Scry is a programmatic search platform that lets an AI agent run structured queries over billions of public web documents in a single operation. Instead of returning a list of links, it executes a bounded query—such as a phrase match, time window, join, or vector similarity search—across a live, queryable index of roughly 165 billion rows, and returns results with each source's own identifiers and provenance fields. The data is exposed via an MCP server at https://mcp.scry.io, which agents connect to after a brief setup. The schema is available at /v1/scry/schema, so an agent can discover which relations are available and how fresh each source is.

Scry is designed to answer questions that the public record can address, but that conventional web search handles poorly—queries where the answer depends on cross-referencing many pages, filtering by date or domain, or measuring similarity across embeddings. It is free for users, runs entirely in the cloud with nothing installed on users' machines, and is positioned as public-benefit infrastructure subsidized for independent, noncommercial research. Commercial use, custom sources, and dataset work are available by writing to the team.