Cover Letter Generator

Generate tailored cover letters from a job posting URL or pasted description. Now with user accounts, persistent profiles, and user-scoped RAG for smarter context.

Features

• Account auth: sign up / log in (local, hashed passwords)
• Persistent user profile: name, education, contacts, links, skills, resume text
• Manage Portfolio items (title, URL, skills, description)
• Manage Certifications (title, issuer, date, skills)
• Manage Experiences (role, organization, years, skills, description)
• User-scoped RAG via ChromaDB: pulls relevant links from your own portfolio/certs/experiences
• Input job via URL or paste description; preferences for tone/style/length
• Download generated cover letter(s)

Setup

1. Create a virtual environment (optional but recommended) and install dependencies:

python -m venv .venv; .\.venv\Scripts\Activate.ps1; pip install -r requirements.txt

2. Create a .env file in app/ or project root with your Groq API key:

GROQ_API_KEY=your_key_here
# Optional: override model
# GROQ_MODEL=llama-3.3-70b-versatile

3. (Optional) Enable Google Login

Create OAuth credentials in Google Cloud Console (OAuth client ID):

• Application type: Web application
• Authorized redirect URI (local): http://localhost:8501
• Copy Client ID and Client Secret

Add to your .env:

GOOGLE_CLIENT_ID=your_google_client_id
GOOGLE_CLIENT_SECRET=your_google_client_secret
# If running on a different host/port, set the redirect URI explicitly
# OAUTH_REDIRECT_URI=http://localhost:8501

4. Run the app:

streamlit run app/main.py

Notes

• URL scraping uses LangChain's WebBaseLoader (BeautifulSoup + requests). If a site blocks scraping, paste the job description instead.
• A local SQLite DB is created at app/data/app.db for users and profile data.
• User-scoped RAG indexes your profile, portfolio items, certifications, and experiences into a ChromaDB collection under vectorstore2/.
• If Google login is configured, the sidebar will show a “Continue with Google” button.

Troubleshooting

• If resume text extraction fails, ensure the file is a valid PDF/DOCX/TXT. For DOCX, python-docx is used; for PDF, pypdf.
• If imports are missing, reinstall dependencies: pip install -r requirements.txt.
• If passlib import fails, run pip install -r requirements.txt to install new dependencies.
• On Windows PowerShell, activate the venv with \.\.venv\Scripts\Activate.ps1.

Roadmap

• Add templates gallery and theme-able outputs (DOCX export)
• Add unit tests and CI workflow
• Support additional LLM providers via config
• OAuth login (GitHub/Google) option