research

name	research
description	Systematic multi-source research across web, GitHub, community discussions, and official docs. Use when asked to research best practices, trends, patterns, or find information on any topic. Invoke with a topic to research.

Research Skill

Systematic multi-source research workflow that produces structured, actionable reports. Use when the user asks you to research best practices, trends, patterns, find information about a tool/concept, or needs up-to-date knowledge beyond your training data.

Workflow

Phase 1: Gather & Explore

Search across these sources in order, using your available search tools. Do not skip sources unless instructed.

Step 1 — Web Search:

• Search for best practices for <topic>, <topic> trends 2025 2026, <topic> patterns, <topic> guide
• Extract: consensus recommendations, common approaches, top tools/libraries, current state of the art

Step 2 — GitHub:

• Search for github <topic> awesome-list, github <topic> best-practices, github <topic> style-guide, github <topic>
• Open and read the top 2-3 most starred/active repositories' READMEs
• Also search <topic> vs <alternative> to understand ecosystem choices
• Extract: code patterns, README recommendations, community conventions, real project examples

Step 3 — Community Discussions:

• Search for reddit <topic>, site:news.ycombinator.com <topic>
• Also search <topic> discussion, <topic> experience, <topic> review
• Extract: real-world experiences, common pitfalls, debated topics, version migration stories, hot takes with evidence

Step 4 — Official Documentation:

• From gathered URLs, fetch official docs pages for the most relevant tools/libraries
• Look for: getting started guides, API references, migration guides, deprecation notes
• Extract: current recommendations, version requirements, configuration details

Configuration:

• Default sources: web,github,community,docs
• Can be filtered: --sources web,docs to skip community/GitHub
• User may specify --output-dir for the report file

Phase 2: Cross-Reference & Synthesize

After gathering, analyze the collected information:

1. Compare findings across sources — note where they agree and where they diverge
2. Tag information quality using these labels:
   • [official] — from official documentation
   • [community consensus] — widely agreed upon across multiple sources
   • [debated] — conflicting opinions or approaches
   • [version specific] — tied to a particular version or release
   • [opinion] — individual or minority perspective
3. Prioritize recency — prefer information from the last 2 years
4. Identify genuine best practices vs. popular-but-dated approaches
5. Note any deprecation warnings or security concerns explicitly

Phase 3: Output

Terminal Summary — Print a compact ~10-line summary first:

Research: <Topic>

• <key finding 1> [source type]
• <key finding 2> [source type]
• ...

Full report saved to: ./research-<topic-slug>.md

Full Report — Save to ./research-<topic-slug>.md (or custom path):

# Research: <Topic>
Generated: <date>

## Executive Summary
Brief 3-5 sentence overview synthesizing the most important findings.

## Best Practices
- <Finding with context and source tag>
- <Finding with context and source tag>

## Trends & Patterns
- <Trend description with evidence>
- <Trend description with evidence>

## Real-World Examples
- <Example from GitHub repo / Reddit thread / article>
- <Example from GitHub repo / Reddit thread / article>

## Official Documentation
- <Key docs page> — <takeaway>
- <Key docs page> — <takeaway>

## Community Discussion
- <Notable Reddit/HN thread> — <key sentiment or debate>
- <Notable Reddit/HN thread> — <key sentiment or debate>

## Recommendations
- <Actionable synthesis and guidance>

## Sources
- <URL 1>
- <URL 2>

Usage Examples

Research a topic with all sources:

Research best practices for structuring tmux configurations

Research with limited sources:

Research neovim lua plugin patterns using only web and github sources

Research with custom output dir:

Research rust async patterns, output to ./research-output/

Triggering

This skill activates when the user:

• Asks to "research" something
• Asks to "find best practices for" something
• Asks to "look into" a topic deeply
• Says "research "
• Asks about current trends, patterns, or state of the art for a topic