deep-research

Generates cited multi-source research reports using firecrawl and exa web search MCPs.

Updated Mar 22, 2026
One-click install
npx skills add https://github.com/diazMelgarejo/orama-system --skill deep-research-diazmelgarejo
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: deep-research
Source: https://github.com/diazMelgarejo/orama-system/tree/main/.cursor/.agents/skills/deep-research
Command: npx skills add https://github.com/diazMelgarejo/orama-system --skill deep-research-diazmelgarejo

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Answering complex questions requires gathering, reading, and synthesizing information from many web sources, which is slow and error-prone when done manually. This Skill automates multi-source web research and delivers structured reports where every claim is backed by a cited source. ## Core Features & Use Cases - Multi-Source Search: Breaks a topic into 3-5 sub-questions and searches each with firecrawl and exa MCP tools, targeting 15-30 unique sources. - Deep Source Reading: Fetches full content of key sources via firecrawl_scrape or crawling_exa instead of relying on search snippets. - Cited Report Generation: Produces a structured report with executive summary, themed sections, inline citations, key takeaways, and a source list. - Parallel Research: Supports launching parallel subagents to research sub-questions concurrently for broad topics. - Use Case: Ask for a deep dive into the competitive landscape for AI code editors and receive a report with sourced findings, market context, and clearly labeled confidence levels. ## Quick Start Ask the agent to research the current state of nuclear fusion energy and deliver a cited report.

Frequently Asked Questions about deep-research

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

FAQPage Schema
How do I do deep web research with cited sources using AI?

Break the topic into 3-5 sub-questions, search each with firecrawl or exa MCP tools, read the top sources in full, then synthesize findings into a report with inline citations and a source list. This Skill automates that entire workflow.

What MCP tools are needed for multi-source web research?

At least one of firecrawl (firecrawl_search, firecrawl_scrape, firecrawl_crawl) or exa (web_search_exa, web_search_advanced_exa, crawling_exa) is required. Using both together gives the best source coverage.

How do I configure firecrawl or exa for research?

Configure the MCP servers in ~/.claude.json or ~/.codex/config.toml with your API credentials. Once configured, the Skill can call their search and crawl tools directly during research.

Can research tasks be parallelized across multiple agents?

Yes. For broad topics, the Skill uses Claude Code's Task tool to launch parallel research agents, each handling different sub-questions. The main session then synthesizes their findings into the final report.

How does the research report handle unverified or missing information?

Claims supported by only one source are flagged as unverified, and sub-questions with insufficient data are explicitly acknowledged as gaps. Estimates and opinions are labeled separately from facts.