Half a Day Finding and Verifying Sources? /deep-research

When researching a new topic, half a day flies by as you open dozens of tabs, compare sources, and verify "can I trust this info." /deep-research is a command that hands all this legwork off to AI. Not a simple search — it gathers multiple sources, cross-verifies them, and organizes a cited report. One question sentence is all it takes.

Definition

/deep-research is a command that broadly researches and cross-verifies one question on the web, organizing it into a cited report.

While a regular web search tosses you a list of links, this goes one step further. It reads multiple sources itself, contrasts and filters conflicting content (cross-verification), and turns it into organized writing. Each claim comes with a citation of where it came from, so you can verify reliability yourself. That's why it's named "deep research."

How to use it (by difficulty)

Basics — research a topic

Append the topic you want to research after the command.

/deep-research 2026 AI coding tool market trends

Then Claude Code starts searching and gathering multiple sources. While research progresses, the process of which sources it's reading is shown, and when done, an organized report appears on screen with citation links. If the research scope is broad, it can take a bit of time.

Applied — comparison research

/deep-research compare the pricing policies of 3 competitors

It's especially strong on topics comparing multiple subjects. It gathers each subject's info and contrasts it in tables or by item, and citations are attached so you can immediately verify "is this price info correct." The more verification matters in the research, the more it shines.

Advanced — chain into blog references

/deep-research vibe coding trends

First research the topic as above, then pass the results to the writing stage.

write a draft from the research I just did with seo-blog-writer

You connect the research report directly as input to the writing skill. Research → writing flows as one, eliminating the hassle of finding sources separately and writing separately.

Common pitfall — always skim the sources

Even though /deep-research goes through cross-verification, it's an AI-made report, so occasionally inaccurate content can slip in. That's why it attaches sources. Build the habit of clicking the attached source link once to verify important numbers or claims. View the report not as a "final version to trust as-is" but as a "first-pass research that even comes with sources," and you can use it most safely and efficiently.

A real case

When I set "vibe coding" as a new blog topic, I handed all the reference research off to /deep-research. Without having to search and compare scattered sources myself, an organized version with sources came out at once. Passing that result straight to the draft-writing skill shrank research-to-draft from a half-day job to a one-hour job.

Use it like this too

• Comparison research: Strong on comparison topics needing verification, like competitors, products, options.
• Research → document link: Chain research results directly into a report, PPT, or blog draft.
• Blog references: Hand off the entire source research for a topic and chain it into the writing stage.
• Sources included: Every claim comes with a citation so you can verify reliability yourself.

Tip: The more specific your question, the more accurate the report. Rather than "AI coding tools," narrow the scope and time, like "Q1 2026 domestic AI coding tool market share."

Wrap-up

The key is one sentence. Instead of legwork, finish research with one question sentence. /deep-research does searching, comparing, verifying, and organizing at once, and even attaches sources. Just pair it with the habit of verifying sources once for important info and it's perfect. Spend the saved research time on your core work.

Based on: Claude Code v2.1.154 (2026.05)