Multi-Agent Job Application System

A powerful, automated Multi-Agent Job Application workflow powered by CrewAI and GCP Vertex AI (Gemini). This system streamlines your job hunting by taking a job URL, scraping the requirements, evaluating your fit, automatically tailoring your resume, drafting a personalized cover letter, and even sending out the application via email on your behalf—all with Human-in-the-Loop (HITL) approval.

🚀 Features

• Automated Web Scraping: Uses Jina Reader API to cleanly extract markdown from job URLs, bypassing complex DOM structures.
• Smart Fit Evaluation: Evaluates the scraped job description against your base resume to determine a GO/NO-GO decision.
• Resume Tailoring: If it's a good fit, the AI tailors a single-column markdown resume (EU compliant) and automatically converts it to a professional PDF using fpdf2.
• Cover Letter Generation: Drafts a highly personalized cover letter based on the job requirements and your background.
• Human-in-the-Loop (HITL) Safety: Generates the materials and saves them in a local "Dry Run" output folder organized by company name for your review before taking action.
• Automated Email Dispatch: Approving the Dry Run prompts the system to automatically email your tailored resume and cover letter directly to the recruiter.

🧰 Tech Stack & Tools

• Language: Python 3.9+
• Agent Orchestration: CrewAI
• LLM Provider: Google Vertex AI (Gemini 3.1 Flash Lite) via LiteLLM
• Extraction: Jina Reader API
• PDF Generation: fpdf2
• Testing: pytest (100% Core Code Coverage)

🛠️ Prerequisites

Before you start, ensure you have the following installed:

1. Python 3.9+
2. API Key for Gemini (Google AI Studio / Vertex AI).
3. An App Password for your SMTP email provider (e.g., Gmail) to dispatch applications.

💻 Installation

1. Clone or Download the Project:

   git clone <repository_url>
   cd "MultiAgent Job Apply"

2. Create and Activate a Virtual Environment (Optional but Recommended):

   python3 -m venv venv
   source venv/bin/activate

3. Install Dependencies:

   pip install -r requirements.txt

⚙️ Configuration

1. Environment Variables: Rename .env.example to .env and fill in your details:

   # Google AI Studio / Vertex AI settings
   GEMINI_API_KEY="your_gemini_api_key_here"
   
   # SMTP Email Settings (for sending applications)
   SMTP_SERVER="smtp.gmail.com"
   SMTP_PORT="465"
   SENDER_EMAIL="your_email@gmail.com"
   SENDER_APP_PASSWORD="your_app_password_here"
   
   # Execution Mode
   DRY_RUN=True

2. Prepare Your Resume: There is a file named base_resume.md in the root directory. Paste your standard, comprehensive resume into this file using Markdown format. The AI will use this as the master reference when tailoring applications.

▶️ Usage

Run the main application script:

python main.py

The Workflow:

1. Input Information: The script will prompt you for:
   • Target Job URL
   • Company Name (used to organize your output files)
   • Recruiter Email (if known; otherwise you can use a test email or leave it out if the system is configured to just generate docs)
2. Phase 1 (Scrape & Evaluate): The agents will scrape the URL and decide if it's a "GO" or "NO-GO" based on your base_resume.md.
3. Phase 2 (Tailor & Draft): If "GO", the system tailors the resume, drafts an email/cover letter, and saves them locally.
4. Review (HITL): Go to the generated folder output/<Company_Name> to review your tailored_resume.pdf and email_dry_run.txt.
5. Dispatch: The console will ask if you want to proceed. Type yes to send the email automatically, or no to abort.

🗂️ Project Structure

• main.py: The entry point that orchestrates the CrewAI process.
• agents.py: Defines the personalized AI agents (Scraper, Evaluator, Tailor, Writer, Executor).
• tasks.py: Outlines the specific tasks assigned to each agent.
• tools/: Contains custom Python tools for PDF generation, sending emails, and scraping.
• output/: Auto-generated folder where tailored resumes and draft emails are stored per company.
• requirements.txt: Python package dependencies.
• base_resume.md: Your master resume in Markdown format.

🤝 Contributing

Feel free to submit issues, fork the repository, and send pull requests.

📝 License

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