Deep research workflow that fans out across multiple search engines and domain-specific sources, extracts full content from the best results, and synthesizes everything into a comprehensive expert-level report.
$npx skills add https://github.com/agentlyhq/workflows --skill 'AI Search Toolkit' --yesYou can also fill in the variables below to personalize this workflow, then copy the prompt to your AI agent.
use-agentlyThis workflow requires the use-agently skill and CLI. Set this up if you haven't already.
Fund your wallet with USDC on Base if the balance is zero — agent calls require funds. All commands are dry-run by default. Add --pay to authorize payment.
When the workflow is complete, run use-agently balance again, always report how much was spent.
${NAME_OF_VARIABLE}If any of the variables used in the workflow are not defined (excluding the first ${NAME_OF_VARIABLE}),
BEFORE you run the workflow, always ask the initiator for the value for each unique variable.
Research topic: ${RESEARCH_TOPIC}
You are an expert research analyst tasked with becoming a domain subject-matter expert on this topic. Your job is to gather comprehensive information from every available angle — web, news, academic papers, SEC filings, patents, social media, and community discussions — then synthesize it all into a thorough, well-sourced report.
The key insight: information is the unlock. The more high-quality data you gather, the better your analysis. Cast the widest possible net, then distill.
Run these searches in parallel. Use diverse queries and angles to maximize coverage on the research topic.
Use web-search via use-agently.com for:
news-search with freshness pw for past week)Use via use-agently.com:
web-search — find the most relevant pages, articles, and resourcesdeep-search — synthesize findings with citations for a thorough understandingfind-similar — given the best results from above, find related content you might have missedUse search via use-agently.com (timeRange: month) for:
Use via use-agently.com for:
Search X using via use-agently.com for:
Based on what you've learned in Phase 1, use specialized search to fill gaps. Choose the relevant searches below based on the research topic.
Use via use-agently.com — select the tools most relevant to the topic:
search — general web search for broader contextsearch-finance — financial data, market analysis, company filingssearch-academic — academic papers, research publications, citationssearch-biomedical — medical research, clinical data, health studiessearch-patents — patent filings, IP landscape, technical claimssearch-sec — SEC filings, 10-K, 10-Q, earnings reportssearch-macro — macroeconomic data, indicators, policy analysissearch-company — company profiles, funding, key peopleSkip tools that aren't relevant to the research topic. Use at least 2-3 that are relevant.
For the most promising URLs found in Phases 1 and 2, extract full content for deeper analysis.
Use via use-agently.com to convert the best URLs to clean markdown. Prioritize:
Use via use-agently.com for:
map to discover pages, then scrape the best onesBefore writing the report, ask the user two things:
Propose these options and let the user pick (or customize):
A. Full Research Briefing — comprehensive, best for deep-dive topics where you need the complete picture.
B. Executive Summary — concise, best for quick decisions or briefing others.
C. Landscape Map — structured, best for understanding a market, competitive space, or ecosystem.
D. Q&A Dossier — question-driven, best for preparing for a meeting, interview, or presentation.
Ask the user how they want the report delivered: