tech-researcher

Tech Researcher

Research technical topics comprehensively by searching multiple sources and applying Fabric patterns for structured analysis. Use this skill when the user wants to:

• Find GitHub repositories for a technology or problem
• Discover blog posts and articles about a topic
• Locate official documentation
• Compare tools, libraries, or frameworks
• Learn about best practices or patterns

Research Process

When researching a topic, follow this structured approach:

1. GitHub Repository Search

Search for relevant repositories using the gh CLI (run via Bash, not a skill):

# Search by topic
gh search repos "<topic>" --limit 10 --sort stars

# Search with language filter
gh search repos "<topic> language:rust" --limit 10 --sort stars

# Search by description
gh search repos "<topic>" --limit 10 --sort updated

For each promising repo, gather details:

# Get repo info
gh repo view <owner>/<repo>

# Check recent activity
gh repo view <owner>/<repo> --json updatedAt,stargazerCount,description

2. Web Search for Articles and Docs

Use WebSearch to find:

• Official documentation
• Blog posts and tutorials
• Comparisons and benchmarks
• Best practices guides

Search queries to try:

• "<topic>" documentation official
• "<topic>" tutorial guide
• "<topic>" vs "<alternative>" comparison
• "<topic>" best practices <current-year>
• "<topic>" examples github

3. Fetch and Summarize Content

Use WebFetch to read promising pages and extract key information:

• Features and capabilities
• Installation/setup instructions
• Pros and cons
• Code examples
• Community activity

4. Apply Fabric Patterns for Analysis

Use Fabric patterns from ~/repos/danielmiessler/Fabric/data/patterns/ to structure your analysis:

For Content Analysis

extract_wisdom - Extract comprehensive insights from articles/docs:

• SUMMARY (25 words)
• IDEAS (20-50 key ideas)
• INSIGHTS (10-20 refined insights)
• QUOTES (notable quotes)
• RECOMMENDATIONS (actionable advice)
• ONE-SENTENCE TAKEAWAY

analyze_paper - Deep analysis of academic papers or technical docs:

• Main arguments and findings
• Methodology evaluation
• Strengths and weaknesses
• Implications

For Comparisons

compare_and_contrast - Structured comparison of tools/libraries:

• Feature comparison matrix
• Pros/cons for each option
• Use case recommendations

For Summarization

summarize_paper - Concise paper/article summaries create_5_sentence_summary - Ultra-brief summaries extract_recommendations - Pull out actionable items

Pattern Usage

To apply a pattern:

1. Read the pattern: cat ~/repos/danielmiessler/Fabric/data/patterns/<pattern>/system.md
2. Follow its IDENTITY, STEPS, and OUTPUT INSTRUCTIONS
3. Process the fetched content through the pattern structure

Output Storage

Always save research results to:

~/repos/scottidler/obsidian/notes/<topic>.md

• <topic>: lowercase-hyphenated slug (e.g., slack-mcp-research, rust-http-clients-research)
• All research notes go to the flat notes/ folder - no subdirectories
• Append -research to the slug to distinguish from ingested notes on the same topic

Output Format

Every research note must have proper YAML frontmatter per the vault schema, followed by the research content.

---
title: "<Human Readable Topic> Research"
date: <YYYY-MM-DD>
type: research
domain: <ai|tech|football|work|writing|music|spanish|knowledge|resources>
origin: assisted
tags:
  - <relevant>
  - <tags>
---

## Top GitHub Repositories

| Repository | Stars | Description |
|------------|-------|-------------|
| owner/repo | 1.2k  | Brief desc  |

## Key Resources

- **Official Docs:** [link]
- **Best Tutorial:** [link]
- **Comparison:** [link]

## Summary

<2-3 paragraph summary of findings>

## Recommendations

<actionable recommendations based on research>

Domain Selection

Pick the domain value that best matches the research topic:

Research about	Domain
AI, LLMs, agents, prompting	ai
Programming, CLI tools, DevOps, infrastructure	tech
Football coaching, schemes, plays	football
Tatari, SRE, work infrastructure	work
Other topics	use best match from domain-values

Research Depth Levels

Adjust depth based on user needs:

Quick Research (default)

• Top 5 GitHub repos by stars
• 2-3 key documentation/article links
• Brief summary

Deep Research (when requested)

• Top 10+ repos with detailed analysis
• Multiple articles and comparisons
• Code examples and patterns
• Community sentiment analysis
• Pros/cons breakdown

Source Priority

Prioritize sources in this order:

1. Official documentation - most authoritative
2. GitHub repos - real implementations and examples
3. Reputable tech blogs - e.g., official company blogs, known authors
4. Stack Overflow - community solutions
5. Tutorial sites - learning resources

Avoiding Noise

Filter out:

• Outdated content (check dates, prefer recent)
• Low-quality SEO articles
• Abandoned repos (no commits in 2+ years, unless stable/complete)
• Paywalled content

Example Research Queries

User Request	Research Approach
"Find a good Rust HTTP client"	Search GitHub for http client language:rust, compare reqwest vs ureq vs hyper
"How does X technology work?"	Find official docs, architecture posts, GitHub readme
"Best practices for Y"	Search for style guides, official recommendations, popular repos' patterns
"Compare A vs B"	Find comparison articles, check GitHub stars/activity, read discussions

Fabric Pattern Reference

Key patterns for research tasks (in ~/repos/danielmiessler/Fabric/data/patterns/):

Pattern	Use Case
extract_wisdom	Comprehensive insight extraction
analyze_paper	Academic/technical paper analysis
analyze_tech_impact	Technology impact assessment
compare_and_contrast	Tool/library comparisons
summarize_paper	Paper summarization
extract_recommendations	Actionable recommendations
extract_references	Source and citation extraction
create_5_sentence_summary	Ultra-brief summaries

Future Capabilities

Not yet implemented:

• YouTube video search and summarization (use extract_wisdom on transcripts)
• Conference talk discovery