DeepRe logo

DeepRe

by hirokidaichi

DeepRe is a Deno-based CLI tool that leverages the Google Gemini AI API to automatically generate in-depth research reports on specified topics. It collects high-quality information through multiple iterative investigations to create comprehensive reports.

View on GitHub

Last updated: N/A

What is DeepRe?

DeepRe is a command-line tool built with Deno that uses the Google Gemini AI API to automate the process of creating in-depth research reports. It iteratively investigates a given topic, automatically evaluates the quality of the information gathered, and outputs a comprehensive report in Markdown format.

How to use DeepRe?

  1. Install Deno and obtain a Google Gemini API key.
  2. Set the GEMINI_API_KEY environment variable or provide the key using the -k option.
  3. Install DeepRe globally using deno task install.
  4. Run DeepRe with the desired research topic: deepre "調査テーマ".
  5. Customize the output directory, Gemini model, and number of iterations using options like -o, -m, and -i.

Key features of DeepRe

  • Iterative Research Process

  • Automatic Evaluation System

  • Gemini API Integration

  • Automatic Research Plan Generation

  • Full Japanese Language Support

  • Markdown Report Output

Use cases of DeepRe

  • Preliminary academic research

  • Market trend and competitor analysis

  • Technology trend research

  • Detailed information gathering on specific topics

FAQ from DeepRe

What is the default Gemini model used by DeepRe?

The default Gemini model is gemini-2.0-flash.

How do I specify the Gemini API key?

You can specify the Gemini API key either by setting the GEMINI_API_KEY environment variable or by using the -k or --api-key option when running DeepRe.

Where are the research reports saved by default?

The research reports are saved in the ./research directory by default. You can change this using the -o or --output-dir option.

How many iterations does DeepRe perform by default?

DeepRe performs 10 iterations by default. You can change this using the -i or --iterations option.

What format are the research reports generated in?

The research reports are generated in Markdown format.