Skip to content

CLAUDE.md

Latest commit

 

History

History
79 lines (55 loc) · 2.79 KB

CLAUDE.md

File metadata and controls

79 lines (55 loc) · 2.79 KB
Raw

CLAUDE.md

This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.

Project Overview

ResumePolisher is a Gradio-based web application providing AI-powered resume enhancement tools. Uses Llama 3.3 70B Instruct via Hugging Face Inference API.

Development Commands

Run the application:

python main.py

Launches a local server at http://localhost:7860.

Install dependencies:

pip install -r requirements.txt

Architecture

Core Pattern: Shared LLM Client

All four tools share a single InferenceClient instance managed via get_hf_client(). This reduces API overhead and improves performance. The client is instantiated once and reused across all tabs.

Component Design

Each tool follows a consistent pattern in main.py:

  1. *_tool() function: Core business logic
  2. _*_tab() function: Gradio UI component construction
  3. Button click handler: Orchestrates function call and error display

The api_key_state Gradio State variable is passed to all tabs and serves as the token source.

Configuration

  • config.py: Centralized model ID configuration
  • .env: Environment variables via python-dotenv
  • USE_ENV_TOKEN flag (line 15): Controls whether HF token comes from UI or environment

Key Functions

Client Management:

  • get_hf_client(token=None): Returns InferenceClient with appropriate token source

Tool Implementations:

  • polish_resume(): Resume content enhancement for specific roles
  • generate_html(): Creates downloadable HTML resumes with photo embedding
  • get_career_advice(): Compares resume to job descriptions
  • generate_cover_letter(): Generates customized cover letters

Error Handling:

  • All tool functions return tuple: (result, error_message)
  • Error messages displayed as HTML blocks using get_error_html()
  • Exceptions caught and returned as user-friendly error strings

Prompt Design Considerations

Each tool uses carefully crafted prompts:

  • Resume Polisher: Role-specific improvements with optional instruction override
  • HTML Generator: Explicit markdown-to-HTML conversion instructions + PHOTO_PLACEHOLDER
  • Career Advisor: Focuses on gap identification between job requirements and resume
  • Cover Letter: Emphasizes aligning real qualifications with job requirements

Security

  • .env contains sensitive tokens (HF_TOKEN)
  • .gitignore prevents token commits
  • API key can be provided via UI (Settings tab) or environment variable
  • Base64-encoded photos stored temporarily for HTML generation

Important Notes

  • Single file entry point (main.py) - no additional module imports
  • All prompts avoid fabricating experience
  • HTML generation uses regex cleanup to remove markdown code fences
  • Temporary HTML files created via tempfile.mkstemp() for downloads