Search large public archives with plain language

Claude Code can translate everyday questions into SQL or vector queries for Scry's read-only data index.

Done with Claude Code· Intermediate

What they did
A developer added Scry, an MCP server at mcp.scry.io, to Claude Code so the agent could inspect the live schema and identify which tables might hold the answer. It then generated a single constrained query, choosing from options like keyword matching, a date range, a join, or vector operations, and sent that request to Scry's servers instead of running it locally.
How it went
The site says a sweep across indexed Hacker News text, totaling 15.5 billion characters, completes in 0.86 seconds. It also lists around 105 billion rows that can be queried, but this example does not show an actual search outcome and does not include a working email-alert example.
Tips
Use it when you want one focused query rather than a broad search process. Pricing is described as very low, and access is free for non-commercial use when capacity is available; results can vary with source freshness.

Try it yourself with Claude Code

I want to ask questions about [public dataset or archive] in plain English. Review the schema for [read-only index or database], pick the tables or fields that are most likely to answer my question, and write one bounded SQL or vector query for: [your question]. Show the query first, then summarize the answer briefly. If I request an email alert, draft the alert conditions and stop until I approve before creating, enabling, sending, or deleting anything.

Fill in the [brackets]. Agents can make mistakes, so review anything before it is sent, paid or deleted.

Read the original on exopriors.com ↗

Get 3 agent uses like this every morning

With the day's top AI news. Free, and one click to unsubscribe.