Research

This is a structured research methodology wrapped around the Tavily API that treats research as a four-phase process: analyze your topic to determine scope, discover expert terminology to escape beginner-level search results, run foundational searches using those expert terms, then explicitly hunt for counter-perspectives. It includes a Deno-based CLI tool for Tavily and templates for vocabulary mapping and stakeholder analysis. The calibration tables linking decision stakes to search depth are genuinely useful. It's opinionated about avoiding confirmation bias and scope creep, which makes it more valuable than generic "how to research" advice. Use this when you're spinning your wheels on a research task or when you need to justify confidence levels in your findings.

skills.sh installs: 382
GitHub stars: 182
last updated: Aug 21, 2026

View on GitHub

Install

$ npx -y skills add jwynia/agent-skills --skill research --agent claude-code

Installs into .claude/skills of the current project.

Files

Research Skill

Tool-assisted research with Tavily integration. Transforms basic questions into comprehensive search strategies using AI-optimized web search.

Setup

This skill includes a bundled Tavily CLI script at scripts/tavily-cli.ts.

Requirements

Deno - Install from https://deno.land
Tavily API Key - Get one at https://tavily.com (free tier available)

Configuration

Set your API key:

export TAVILY_API_KEY="your-key-here"

Create an alias for convenience (add to your shell profile):

# Adjust path to where this skill is installed
alias tavily='deno run --allow-net --allow-env /path/to/skills/research/scripts/tavily-cli.ts'

Or run directly:

deno run --allow-net --allow-env ./scripts/tavily-cli.ts "your query"

Commands below use tavily assuming the alias is configured.

Quick Reference

Common Commands

# Basic search
tavily "your query"

# With AI answer summary
tavily "your query" --answer

# Deep search with more results
tavily "your query" --depth advanced --results 10 --answer

# News/recent content
tavily "your query" --topic news --time week

SKILL.md on GitHub