deep-research-pro

Generate citation-backed research reports with multi-agent retrieval and triangulated verification.

1|Updated Mar 26, 2026
One-click install
npx skills add https://github.com/mediar-ai/deep-research-pro --skill deep-research-pro-mediar-ai
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: deep-research-pro
Source: https://github.com/mediar-ai/deep-research-pro/tree/main
Command: npx skills add https://github.com/mediar-ai/deep-research-pro --skill deep-research-pro-mediar-ai

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

It helps you generate production-grade, comprehensive research reports with multi-agent orchestration and strong write-after-search reliability so you can make decisions based on well-sourced evidence rather than shallow browsing.

Core Features & Use Cases

  • Multi-phase deep research pipeline: Runs a structured 8-phase workflow (scope, plan, retrieve, triangulate, synthesize, critique, refine, and package) to move from question decomposition to final reporting.
  • Multi-agent parallel retrieval with reliability protocol: Uses parallel agent tracks with a strict write-after-search requirement and stuck-agent recovery to improve completeness and reduce failure modes.
  • Triangulated, citation-dense outputs: Cross-verifies major claims across 3+ independent sources and packages results into Markdown, HTML, and PDF with a claims verification table and bibliography.
  • Optional human checkpoints: Supports interactive mode when the user requests an outline/review, otherwise runs autonomously.

Quick Start

Ask for a deep research report by saying something like: “Run deep research on compare X vs Y with citations and a thorough claims verification table, using the standard mode.”

Frequently Asked Questions about deep-research-pro

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I generate a citation-backed research report with cross-source verification?

To generate a citation-backed research report, use a multi-agent parallel retrieval pipeline that decomposes complex questions, triangulates major claims across multiple independent sources, and packages the synthesized findings into Markdown, HTML, and PDF formats with a bibliography.

What is triangulated claim verification in deep research?

Triangulated claim verification in deep research is the process of cross-verifying major claims across three or more independent sources to ensure reliability, producing a claims verification table that validates evidence before final report synthesis.

How do I run multi-agent parallel retrieval for complex topic analysis?

You run multi-agent parallel retrieval by executing a structured multi-phase workflow that includes scoping, planning, retrieving, and synthesizing, utilizing parallel agent tracks with a strict write-after-search requirement to improve completeness.

Can I review the research outline before the full report is generated?

Yes, you can review the research outline by requesting an interactive human checkpoint mode, which allows you to review the outline before the pipeline autonomously continues through retrieval, triangulation, and final packaging.

What is the best way to handle stuck-agent recovery during multi-agent orchestration?

The best way to handle stuck-agent recovery during multi-agent orchestration is to use a reliability protocol within the parallel retrieval pipeline that automatically detects and recovers failed agent tracks to reduce failure modes and ensure complete research.

What output formats are supported for packaged deep research reports?

Packaged deep research reports are supported in Markdown, HTML, and PDF formats, providing a comprehensive briefing that includes a claims verification table and a bibliography for citation-dense outputs.