An AI-powered agent built with Google's Agent Development Kit (ADK) that automatically searches and analyzes Angular's GitHub pull requests for new features.

This agent:

• Fetches open PRs from Angular's GitHub repository filtered by "feat"
• Caches PR data locally to minimize GitHub API calls
• Uses Google's Gemini 2.0 Flash model to analyze and format results

• Python 3.10 or higher
• A Google API key for Gemini (get one at Google AI Studio)
• Internet connection for GitHub scraping

python3 -m venv .venv
source .venv/bin/activate  # On macOS/Linux

pip install -r requirements.txt

The .env file is already present with:

GOOGLE_GENAI_USE_VERTEXAI=FALSE
GOOGLE_API_KEY=your_api_key_here

If the Google ADK CLI is installed, you can run the agent directly:

# Run the agent interactively
adk web

By default, the agent searches:

https://github.com/angular/angular/pulls?q=is%3Apr+feat+is%3Aopen

You can modify this in config.py or pass a custom URL to search_github_repository().

PR data is cached in pr_cache.json in the same directory as config.py. This prevents redundant GitHub requests.

• User Agent: FeatureScoutAngularAgent/1.0
• Timeout: 10 seconds
• Headers: Standard browser-like headers

Modify these in config.py if needed.

1. Fetch PR List: The agent fetches the Angular GitHub PR listing page
2. Parse PRs: Extracts all PR links with data-hovercard-type="pull_request"
3. Filter by "feat": Only processes PRs with "feat" in the title (case-insensitive)
4. Check Cache: Looks up each PR in the local cache
5. Fetch Descriptions: If not cached, fetches the PR page and extracts the first comment body
6. Update Cache: Saves new PR data to pr_cache.json
7. AI Analysis: The agent uses Gemini to analyze and format results