deep-research

Cited research reports from fanned-out web search with adversarial verification.

Score 0(?)OfficialCodeby AnthropicSource
Verified for:

Install

Built into Claude Code. Invoke with /deep-research.

Best for

Any deep, multi-source, fact-checked research request — competitive analysis, technical surveys, market sizing.

About this skill

Multi-source research harness — fan-out web searches, fetch sources, adversarially verify claims, synthesize a cited report.

ResearchWeb SearchSynthesis

Score breakdown

Score breakdown

rubric 1.0
Install count
0/20
Provenance
15/15
GitHub stars
0/15
Recency
10/10
Compatibility
4/10
Documentation depth
6.5/10
Install ergonomics
10/10
License
5/5
Verification freshness
5/5

Composite 0–100 score derived from 9 verifiable signals. See the rubric →

Excel spreadsheets with real formulas, charts, and formatting — not flat CSV exports.

ExcelFormulasCharts
All surfaces

Web scraping, search, and browser automation built to be agent-friendly.

ScrapingSearchWeb
All surfaces

Web search + 36 authoritative data sources. SEC, PubMed, ChEMBL, FRED — agent-grade research.

SearchSECPubMed
All surfaces

Skill FAQ

About deep-research

What is the deep-research skill?

Multi-source research harness — fan-out web searches, fetch sources, adversarially verify claims, synthesize a cited report.

Which agents does deep-research work with?

Claude Code, Antigravity IDE.

How do I install deep-research?

See the install panel on this page for the per-agent install commands.

Is deep-research free to use?

Yes — deep-research is open source (MIT).

Who maintains deep-research?

deep-research is published by Anthropic and was last updated on 2026-05-20.

What is deep-research best for?

Any deep, multi-source, fact-checked research request — competitive analysis, technical surveys, market sizing.