Imported from David-Li0406/meta-skill-evloving (skill-flow/data/skills-refined-autoskill-36k/skillsmp/person-profiler/SKILL.md). Install upstream with npx skills add David-Li0406/meta-skill-evloving --skill person-profiler. Copyright stays with the author.

Person Profiler

This skill performs preliminary internet research on individuals and generates a structured markdown report to enable comprehensive follow-up research.

When to Use

Invoke this skill when the user:

• Says "research [person name]"
• Asks "who is [person name]"
• Requests a profile or background on someone
• Wants preliminary information about an individual

Research Process

1. Initial Web Search: Search for the person's name to gather basic information

2. Follow-up Searches: Based on initial findings, search for:

   • Professional affiliations (companies, organizations)
   • Content they've created (blogs, podcasts, YouTube channels)
   • Media appearances (interviews, articles, podcasts)
   • Recent news or updates

3. Create Directory: Create a directory using the person's name in lowercase with underscores (e.g., "John Doe" → john_doe/)

4. Generate Files: Create two files in the person's directory:

   • profile.md - The research report with four sections
   • search.sh - Bash script with all search queries

Report Structure

The report MUST contain exactly these four sections:

1. Summary

• Who the person is (profession, role, title)
• Background (education, career history, notable achievements)
• Current affiliations (company, organization, institution)
• Areas of expertise or focus
• Notable accomplishments or contributions

2. Related Search Terms

A bulleted list of entities and terms associated with this person that could yield additional information:

• Company names they founded/work for
• Podcasts they host or co-host
• Blogs or substacks they write
• Books they've authored
• Projects or initiatives they lead
• Brands or products they're associated with
• Organizations they're affiliated with

3. Media Presence

Document the types of media where this person appears or creates content:

• Podcasts: List any podcasts they host, co-host, or frequently appear on
• YouTube: Channels they own or videos they appear in
• Written Content: Blogs, substacks, Medium, personal websites
• Social Media: Active platforms (Twitter/X, LinkedIn, etc.)
• News Coverage: Types of publications that cover them
• Speaking: Conferences, talks, interviews they give

4. Google Search Terms for Deep Research

A comprehensive bulleted list of specific Google search queries designed to find news, updates, and content about this person. Include:

• Basic searches: "[person name]"
• Site-specific searches for major publications:
  • "[person name]" site:nytimes.com
  • "[person name]" site:wsj.com
  • "[person name]" site:techcrunch.com
  • "[person name]" site:substack.com
  • etc.
• Company/project searches:
  • "[company name]" site:nytimes.com
  • "[company name]" news
• Content-specific searches:
  • "[person name]" podcast
  • "[person name]" interview
  • "[person name]" youtube
• Time-bound searches:
  • "[person name]" after:2024-01-01
  • "[company name]" after:2024-01-01
• Quote and exact match searches for disambiguation

Output Format

1. Create a directory: Make a directory named {person_name}/ (using underscores, lowercase) in the current working directory
2. Save the profile: Save the report as {person_name}/profile.md
3. Generate search script: Create {person_name}/search.sh containing all search queries from Section 4
4. Make script executable: Run chmod +x {person_name}/search.sh
5. Execute the script: Change into the directory and run the script: cd {person_name} && ./search.sh

Example: If researching "Vaden Masrani", create directory vaden_masrani/ with:

• vaden_masrani/profile.md - The research profile
• vaden_masrani/search.sh - Executable bash script with search queries
• vaden_masrani/results.db - SQLite database with search results (created when script runs)

The script must be executed from within the person's directory so that results.db is created there.

Search Script Format

The search script should:

• Be a zsh script with proper shebang (#!/usr/bin/env zsh)
• Use gum for pretty output
• Call ddgs "[query]" --db results.db for each search term from Section 4
• Be executable (chmod +x)

Example search.sh structure:

#!/usr/bin/env zsh

gum style --border rounded --padding "1 2" --bold "Running Google searches for [Person Name]"

echo ""
gum style --bold "Basic searches"
ddgs "[person name]" --db results.db
ddgs "[person name] latest news" --db results.db
ddgs "[person name] 2024" --db results.db

echo ""
gum style --bold "Site-specific searches"
ddgs "[person name] site:nytimes.com" --db results.db
ddgs "[person name] site:wsj.com" --db results.db
ddgs "[person name] site:techcrunch.com" --db results.db
ddgs "[company name] site:nytimes.com" --db results.db

echo ""
gum style --bold "Content-type searches"
ddgs "[person name] podcast interview" --db results.db
ddgs "[person name] youtube" --db results.db
ddgs "[person name] blog post" --db results.db

echo ""
gum style --bold "Time-bound searches"
ddgs "[person name] after:2024-01-01" --db results.db
ddgs "[company name] after:2024-01-01" --db results.db

echo ""
gum style --foreground 2 "✓ All searches complete! Results saved to results.db"

Example Report Template

# [Person Name] - Preliminary Research Profile

## 1. Summary

[2-4 paragraph overview covering who they are, background, affiliations, and expertise]

## 2. Related Search Terms

- [Company name]
- [Podcast name]
- [Blog/Substack name]
- [Project or initiative]
- [Organization name]

## 3. Media Presence

**Podcasts**
- [Podcast they host or appear on]

**YouTube**
- [Channel or notable videos]

**Written Content**
- [Blog, Substack, or website]

**Social Media**
- [Active platforms]

**News Coverage**
- [Types of publications that cover them]

**Speaking/Conferences**
- [Notable talks or events]

## 4. Google Search Terms for Deep Research

Basic searches:
- "[person name]"
- "[person name]" latest news
- "[person name]" 2024
- "[person name]" 2025

Site-specific searches:
- "[person name]" site:nytimes.com
- "[person name]" site:wsj.com
- "[person name]" site:techcrunch.com
- "[person name]" site:substack.com
- "[person name]" site:medium.com
- "[company name]" site:nytimes.com
- "[podcast name]" site:spotify.com

Content-type searches:
- "[person name]" podcast interview
- "[person name]" youtube
- "[person name]" blog post
- "[person name]" conference talk

Time-bound searches:
- "[person name]" after:2024-01-01
- "[company name]" after:2024-01-01
- "[person name]" before:2024-01-01 after:2023-01-01

Combined searches:
- "[person name]" AND "[topic]"
- "[company name]" AND "funding"
- "[person name]" AND "controversy"

Best Practices

1. Use multiple search queries: Don't rely on a single search. Try variations of the person's name and related terms.

2. Verify information: Cross-reference facts from multiple sources when possible.

3. Note the date: Include when the profile was generated, as information may become outdated.

4. Be thorough with search terms: Think creatively about what sites might have information (industry-specific publications, academic databases, social media platforms).

5. Consider disambiguation: If the name is common, add context to search terms (e.g., "[person name]" AND "[company]" OR "[field]").

6. Respect privacy: Focus on publicly available information. Don't speculate about private matters.

7. Generate comprehensive search script: Include ALL search queries from Section 4 in the search.sh script. Each query should be a separate ddgs call. Group related queries with gum style headers for better organization.

8. Automatically execute searches: After creating and making the script executable, change into the person's directory and run it immediately to gather all search results into results.db within that directory.

Notes

• This is a preliminary research tool. The report is designed to enable deeper, more focused research in a second pass.
• The search script is automatically executed after creation, saving all results to a SQLite database ({person_name}/results.db) for easy querying and analysis.
• You can re-run the search script later by executing: cd {person_name} && ./search.sh
• All search results are stored in {person_name}/results.db which can be queried using sqlite3 or other database tools.
• Adapt the depth of research based on how prominent the person is (public figures will have more information available).