Search Agent Using Google ADK

A powerful search agent built with the Google Agent Development Kit (ADK) that performs web searches and content extraction.

Overview

This project implements a sophisticated, autonomous search agent powered by Google's Agent Development Kit (ADK) and the Gemini 2.5 Flash model. It's designed to be a multi-platform research tool that leverages AI to intelligently search, extract, and organize information from the web.

What Makes It Special

🔗 Multi-Source Search: Instead of relying on a single search engine, the agent can search across:

• Google
• GitHub repositories
• Reddit communities and discussions
• Stack Overflow questions and answers
• Wikipedia articles

Technology Stack

• Python 3.7+
• Google ADK - Agent Development Kit for building AI agents
• BeautifulSoup4 - HTML parsing and content extraction
• Requests - HTTP library for web requests

Installation

1. Clone the repository:

git clone <repository-url>
cd Search-Agent-Using-Google-ADK

2. Create Virtual Environment:

python -m venv venv

3. Set up environment variables: Go to the search-agent folder, then create a .env file based on .env.example and add your Google API credentials:

GOOGLE_GENAI_USE_VERTEXAI=0
GOOGLE_API_KEY=your_google_genai_api_key
GOOGLE_CSE_ID=your_google_custom_search_engine_id
SEARCH_API_KEY=your_google_search_api_key

• GOOGLE_GENAI_USE_VERTEXAI: Set to 0 to use standard Google AI API (or 1 for Vertex AI)
• GOOGLE_API_KEY: Your Google Gemini API key. Get it from Google AI Studio
• GOOGLE_CSE_ID: Your Google Custom Search Engine ID. Create a custom search engine at Google Custom Search
• SEARCH_API_KEY: Your Google Search API key for programmatic search access. Get it from Google Cloud Console

4. Install dependencies:

pip install -r requirements.txt

5. Run the App:

adk web

6. Open the URL from the terminal

Dependencies

The project requires the following Python packages:

• requests>=2.25.0 - HTTP requests library
• beautifulsoup4>=4.9.0 - HTML/XML parsing
• google-adk - Google Agent Development Kit

Project Structure

Search-Agent-Using-Google-ADK/
├── README.md                 # This file
├── requirements.txt          # Python dependencies
└── search-agent/
    ├── __init__.py          # Package initialization
    ├── agent.py             # Main agent implementation
    ├── __pycache__/         # Python cache directory
    └── data/                # Storage for search results and content

Tools & Functions

The search agent includes the following tools that can be used autonomously:

search_google(query: str) -> dict

Performs a web search using Google Custom Search API (CSE).

• Input: query - Search query string
• Output: Dictionary with query, source, results, and total count
• Uses: SEARCH_API_KEY and GOOGLE_CSE_ID environment variables
• Returns per result: title, URL, snippet

search_github(query: str, sort: str) -> dict

Searches GitHub for repositories.

• Input:
  • query - Search query string
  • sort - Sort by (default: "stars", options: stars, forks, updated)
• Output: Dictionary with query, total count, and repository results
• Returns per repo: name, URL, description, stars, language, updated_at

search_reddit(query: str, sort: str) -> dict

Searches Reddit for posts across all subreddits.

• Input:
  • query - Search query string
  • sort - Sort by (default: "relevance")
• Output: Dictionary with query and post results
• Returns per post: title, URL, subreddit, score, num_comments, author, selftext (truncated to 500 chars)

search_reddit_r(subreddit: str, query: str, sort: str) -> dict

Searches within a specific subreddit.

• Input:
  • subreddit - Subreddit name (without r/)
  • query - Search query string
  • sort - Sort by (default: "relevance")
• Output: Dictionary with query, subreddit, and post results
• Returns per post: title, URL, score, num_comments, author, selftext (truncated to 500 chars)

search_stackoverflow(query: str, sort: str) -> dict

Searches Stack Overflow for questions.

• Input:
  • query - Search query string
  • sort - Sort by (default: "relevance")
• Output: Dictionary with query, total count, and question results
• Returns per question: title, URL, score, answer_count, is_answered, tags, body (truncated to 300 chars)

search_wikipedia(query: str) -> dict

Searches Wikipedia for articles.

• Input: query - Search query string
• Output: Dictionary with query and article results
• Returns per article: title, URL, snippet, content (truncated to 1000 chars)

fetch_url_content(url: str) -> dict

Extracts and parses content from a URL.

• Input: url - URL string
• Output: Dictionary with URL, title, content, and status code

write_to_file(file_name: str, content: str) -> dict

Saves content to a file in the data directory.

• Input:
  • file_name - Name for the file
  • content - Content to write
• Output: Status dictionary with file path, success status, and content size

get_current_data_files() -> dict

Lists all files in the data directory.

• Output: Dictionary with success status and list of file names

read_from_file(file_name: str) -> dict

Reads content from a file in the data directory.

• Input: file_name - Name of the file to read
• Output: Dictionary with file path, content, size, and status

License

This project is open source and available under the MIT License.

Contributing

Contributions are welcome! Please feel free to submit pull requests or open issues for bugs and feature requests.