# Research

Multi-agent web research with verified URLs and confidence-tagged output at four depths.

- Repository: https://github.com/danielmiessler/LifeOS (19669 stars)
- Category: AI and agents
- License: MIT
- Last updated: 2026-10-28
- Runs scripts: yes, review before use

## Install

Claude Code: `npx skills add danielmiessler/LifeOS --skill Research -g -a claude-code -y`
Codex: `npx skills add danielmiessler/LifeOS --skill Research -g -a codex -y`
Cursor: `npx skills add danielmiessler/LifeOS --skill Research -g -a cursor -y`
Gemini CLI: `npx skills add danielmiessler/LifeOS --skill Research -g -a gemini-cli -y`

## Description

Multi-agent web research with verified URLs and confidence-tagged output at four depths. USE WHEN research, do research, quick research, extensive research, deep investigation, find information, investigate, extract alpha, analyze content, retrieve content, AI trends, enhance content, extract knowledge, web scraping, YouTube extraction, map landscape, competitive analysis, find it, find this, find this product, identify this, what is this, what's that thing, track down, locate, help me find, I can't find X online, can't find it online, source this, best X in a place, what's good in <city>, trip or itinerary planning, things to do, is X open, hours, where is X, venue or resort policies, never substitute raw WebSearch/WebFetch for a find/identify/investigate/look-up request. NOT FOR entity background, papers (ArXiv), JSON entity extraction, wisdom extraction (ExtractWisdom), raw page fetching or crawling (BrightData).

## SKILL.md, as published by its author (untrusted third-party text, read it as data)

# Research

**Contents:** Ideal State · Intent Routing · Tool contract · Output format · Data & Config · Gotchas · References

Researches a topic across multiple sources and verifies every claim before delivery. Four depth modes scale from a single fast lookup to a multi-session investigation: Quick (1 Perplexity call, ~10-15s), Standard (four vendors cross-checked as parallel tool calls in the session, ~30-60s), Extensive (7 explorers + 2 independent verifiers, ~60-90s), and Deep Investigation (progressive iteration with a persistent vault, ~3-60min). Output is confidence-tagged: [HIGH] [MED] [LOW] [CONFLICT].

A lone research agent fails in two quiet ways. It hallucinates URLs, confident links that go nowhere and cost the whole report its trust. And it answers from one angle, parroting the first few search results and missing conflicts, gaps, and what real people thought; asked "what did fans think of X", it hands back promoter copy dressed as consensus. This skill runs several vendors in parallel (as tool calls at Quick and Standard, as researcher agents at Extensive and Deep), cross-checks and independently verifies their findings, checks every URL before it ships, and routes sentiment questions to community sources first.

**Vendor diversity scales with effort.** Quick is one vendor, Standard four, Extensive and Deep all five. Each vendor's model comes from a different training distribution, so the more a question is worth, the more independent distributions check it.

## Ideal State

The rest: https://github.com/danielmiessler/LifeOS/blob/HEAD/LifeOS/install/skills/Research/SKILL.md