Research

Fact-check a rumor with deep research and turn it into a published article

The author treated a viral industry rumor as a hypothesis, using a deep-research model to trace the original sourcing, weigh each outlet's credibility, separate amplification from independent confirmation, and assign probabilities to each claim. A custom Claude Code skill then turned the same research into a Substack article, structured data, infographics and social posts.

Tool

GPT-5.6 Sol deep research and Claude Code

Source

@BretKerr on X

Source date

Aug 9, 2026

Credit: this use case was shared by @BretKerr on X. Summary written by Kitbox; see the original for full details.

Try it yourself

Our own starter prompt. Fill in the brackets, then paste it into GPT-5.6 or your agent of choice.

Treat this claim as an unproven hypothesis: [claim]. Find where it first appeared, who repeated it, and which reports confirm it independently versus just re-sharing it. Rate each source's track record and give me a probability for each part of the claim with your reasoning. Then outline a short article that presents only what the evidence supports, with every fact linked to its source.