Searcher — Online Research Tool

Deep-research a person or business using public sources, then get an AI-synthesized report from Claude.

What It Does

Searcher gathers data from multiple public sources — no API keys required for most of them — and feeds everything to Claude AI to produce a structured research report.

Sources searched:

Source	Applies to	Requires
DuckDuckGo web search	Both	Nothing
DuckDuckGo news search	Both	Nothing
Wikipedia	Both	Nothing
Hacker News (Algolia)	Both	Nothing
Reddit	Both	Nothing
SEC EDGAR filings	Both	Nothing
GitHub users & repos	Both	Nothing
WHOIS / domain info	Business	Nothing
Wayback Machine snapshots	Business	Nothing
CourtListener court records	Both	COURTLISTENER_TOKEN (free account)
AI synthesis report	Both	ANTHROPIC_API_KEY

Installation

Requirements: Python 3.10+

git clone <repo-url>
cd Searcher
pip install -r requirements.txt

Setup

For AI synthesis (the main feature), set your Anthropic API key:

export ANTHROPIC_API_KEY=sk-ant-...

Optional — for court record searches via CourtListener (free account):

export COURTLISTENER_TOKEN=your-token-here

Usage

python main.py TARGET [OPTIONS]

Examples

# Research a person
python main.py "Elon Musk"

# Research a business
python main.py "OpenAI" --type business

# Deep research with report saved to file
python main.py "Jane Smith" --depth deep --output report.md

# Business with explicit domain for WHOIS
python main.py "Acme Corp" --type business --domain acme.com

# Skip AI synthesis (no API key needed)
python main.py "SpaceX" --no-ai

Options

Option	Short	Values	Default	Description
--type	-t	person, business, auto	auto	Type of target. auto detects from the name.
--depth	-d	quick, standard, deep	standard	Controls how many search queries are run.
--domain		any domain	auto-guessed	Explicit domain for WHOIS (business targets).
--output	-o	file path	none	Save the AI report to .md or .txt.
--no-ai		flag	off	Skip Claude AI; show only the raw gathered data.

Depth levels

Level	Results per query	Best for
quick	5	Fast overview
standard	8	Balanced (default)
deep	15	Thorough research

Output

The tool prints a live progress display as it gathers data, a summary table of what was found, then streams the AI-generated report in real time.

Report sections (where data exists):

• Overview
• Background & History
• Professional / Business Profile
• Online Presence
• News & Media Coverage
• Key Connections & Associations
• Key Findings

Use --output report.md to save the report to a file.

Notes

• All sources are publicly available information only. No login-walled or paywalled data is accessed.
• Reddit and the Wayback Machine can occasionally rate-limit automated requests — the tool handles this gracefully and continues with whatever it could fetch.
• CourtListener changed their policy and now requires a free account token. Sign up at courtlistener.com to enable it.
• GitHub's unauthenticated API allows 60 requests/hour. If you hit the limit, results will simply be empty for that source.
• The AI report is generated by claude-opus-4-7 with adaptive thinking enabled.