LangChain Open Deep Research Skill

This skill utilizes the LangChain Open Deep Research framework to perform iterative web research with reflection and knowledge gap identification, producing comprehensive reports with citations.

Setup

1. Dependencies: Requires the open-deep-research package and LangGraph.

   bashCopy

   pip install open-deep-research langgraph-cli python-dotenv
   

2. API Key Configuration: Requires API keys for an LLM and a search provider.

   bashCopy

   # Set up your API keys
   echo "# LLM Configuration" >> .env
   echo "OPENAI_API_KEY=your_openai_key" >> .env
   echo "# Search Configuration" >> .env
   echo "TAVILY_API_KEY=your_tavily_key" >> .env
   if [ -f .gitignore ] && ! grep -q ".env" .gitignore; then echo ".env" >> .gitignore; fi
   echo "API keys saved to .env."
   

Usage

Use the scripts/research.py script to run a research task.

Command

python3 scripts/research.py --query "<research_query>" [--max-iterations <N>]

Parameters

• --query (Required): The research question or topic.
• --max-iterations (Optional): Maximum number of research iterations (default: 3).
• --output (Optional): Output file path for the final report (default: stdout).

Example

python3 scripts/research.py --query "What are the latest developments in quantum computing error correction?" --max-iterations 4 --output report.md

Output

The script outputs a comprehensive research report with:

• Iterative search findings
• Knowledge gap analysis
• Final synthesized report with citations
• Source list

Features

• Iterative Research: Performs multiple search cycles, reflecting on gaps
• Configurable Models: Supports OpenAI, Anthropic, Ollama, and other LLM providers
• Multiple Search Engines: Tavily (default), Brave, DuckDuckGo, SerpAPI
• Citation Tracking: All findings include source references