Perform preliminary internet research on a person and generate a structured report. Use when the user asks to research, profile, or get information about a specific person by name.
This skill performs preliminary internet research on individuals and generates a structured markdown report to enable comprehensive follow-up research.
Invoke this skill when the user:
Initial Web Search: Search for the person's name to gather basic information
Follow-up Searches: Based on initial findings, search for:
Create Directory: Create a directory using the person's name in lowercase with underscores (e.g., "John Doe" → john_doe/)
Generate Files: Create two files in the person's directory:
profile.md - The research report with four sectionssearch.sh - Bash script with all search queriesThe report MUST contain exactly these four sections:
A bulleted list of entities and terms associated with this person that could yield additional information:
Document the types of media where this person appears or creates content:
A comprehensive bulleted list of specific Google search queries designed to find news, updates, and content about this person. Include:
"[person name]""[person name]" site:nytimes.com"[person name]" site:wsj.com"[person name]" site:techcrunch.com"[person name]" site:substack.com"[company name]" site:nytimes.com"[company name]" news"[person name]" podcast"[person name]" interview"[person name]" youtube"[person name]" after:2024-01-01"[company name]" after:2024-01-01{person_name}/ (using underscores, lowercase) in the current working directory{person_name}/profile.md{person_name}/search.sh containing all search queries from Section 4chmod +x {person_name}/search.shcd {person_name} && ./search.shExample: If researching "Vaden Masrani", create directory vaden_masrani/ with:
vaden_masrani/profile.md - The research profilevaden_masrani/search.sh - Executable bash script with search queriesvaden_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.
The search script should:
#!/usr/bin/env zsh)gum for pretty outputddgs "[query]" --db results.db for each search term from Section 4Example 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"
# [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"
Use multiple search queries: Don't rely on a single search. Try variations of the person's name and related terms.
Verify information: Cross-reference facts from multiple sources when possible.
Note the date: Include when the profile was generated, as information may become outdated.
Be thorough with search terms: Think creatively about what sites might have information (industry-specific publications, academic databases, social media platforms).
Consider disambiguation: If the name is common, add context to search terms (e.g., "[person name]" AND "[company]" OR "[field]").
Respect privacy: Focus on publicly available information. Don't speculate about private matters.
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.
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.
{person_name}/results.db) for easy querying and analysis.cd {person_name} && ./search.sh{person_name}/results.db which can be queried using sqlite3 or other database tools.