skills.rest
ruban-s
ruban-sCOMMUNITY
Explore all skills by this author →
WebsiteGitHub

Research

Runs multi-agent web research with URL verification and confidence-tagged findings across four depth modes.

Updated Sep 3, 2026
Search Tags:#research#web search#multi-agent#sentiment analysis#competitive analysis#deep investigation#fact verification
One-click install
npx skills add https://github.com/ruban-s/DevOS --skill research-ruban-s
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: Research
Source: https://github.com/ruban-s/DevOS/tree/main/skills/Research
Command: npx skills add https://github.com/ruban-s/DevOS --skill research-ruban-s

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Single-agent research hallucinates URLs and answers from one angle, parroting the first few search results while missing conflicts, gaps, and real community sentiment. This skill runs multiple research agents in parallel, cross-checks their findings, verifies every URL before delivery, and routes sentiment questions to community sources like Reddit and YouTube instead of recap journalism. ## Core Features & Use Cases - Four depth modes: Quick (1 agent, ~10-15s), Standard (3 agents cross-checked, ~30-60s), Extensive (7 explorers + 2 verifiers, ~60-90s), and Deep Investigation (iterative with a persistent knowledge vault, ~3-60min). - Three-layer verification: agent self-verification, cross-check synthesis, and independent verifier agents, producing confidence-tagged output ([HIGH]/[MED]/[LOW]/[CONFLICT]). - Source routing protocol: sentiment questions route to Reddit JSON API, X API v2, and YouTube Data API before web search, following an API-first, scraper-second, search-last cascade. - Use Case: Ask "do research on AI agent frameworks" and get a synthesized report from Claude, Gemini, and Perplexity agents with verified citations and conflict flags in under a minute. ## Quick Start Ask the agent to research a topic, for example by saying "do research on the current state of AI agent frameworks", and it will run the Standard multi-agent workflow with verified sources.

Frequently Asked Questions about Research

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

FAQPage Schema
How do I run multi-agent web research with verified sources?▼

Say "research" plus your topic and the skill launches the Standard mode with three parallel agents (Claude, Gemini, Perplexity) whose findings are cross-checked. Every URL is verified via WebFetch or curl before delivery, and findings carry [HIGH]/[MED]/[LOW]/[CONFLICT] confidence tags.

What research depth modes are available and when should I use each?▼

Four modes exist: Quick (1 agent, ~10-15s) for fast lookups, Standard (3 agents, ~30-60s) as the default, Extensive (7 explorers + 2 verifiers, ~60-90s) for comprehensive coverage, and Deep Investigation (~3-60min) for iterative landscape mapping with a persistent vault.

How does the skill handle fan sentiment or community opinion questions?▼

Sentiment signals trigger source routing to community platforms first: Reddit JSON API, X API v2 recent-search, and YouTube Data API. Web search is used last for context only, since press recaps often invent consensus rather than reflect what people actually said.

Can this skill be used for due diligence or background checks on people and companies?▼

No. The skill explicitly routes entity deep-background, due diligence, and vetting requests to a dedicated OSINT/entity-investigation skill. It also excludes academic papers (use ArXiv) and JSON entity extraction.

Why do research results sometimes mark findings as CONFLICT or LOW confidence?▼

CONFLICT appears when explorer agents and independent verifier agents disagree on a claim, and both sides are reported. LOW marks findings that could not be independently verified, which happens when sources are thin or URLs fail verification.

What happens when a research request includes an X or Twitter URL?▼

A precheck scans for x.com or twitter.com links before spawning agents, since X blocks generic fetching. The skill then tries a dedicated X reader skill, the X API v2 with X_BEARER_TOKEN, or Apify/BrightData scrapers, and asks you to paste the post text if none work.

Related Skills

Explore More Education & Research Skills→
aiyinluyaaiyinluya
2

sn-search-social-en

Search Reddit, Twitter/X, and YouTube for English-language content.

Community
Basic
sharkitect-solutionssharkitect-solutions

content-research-writer

Create research-backed articles with source discovery and citation management.

Community
Advanced
GratefulJinx77GratefulJinx77

osint

Conduct structured OSINT investigations on people and companies using public sources.

Community
Advanced
tanaertanaer

academic-research

Search scholarly databases, download papers, and extract key information.

Community
Intermediate