Job Search Assistant
stepmn818/job-search-assistant
Análise de fit CV vs vaga com score ponderado e recomendações acionáveis
Ver no GitHubSkill que analisa o alinhamento entre seu CV e uma descrição de vaga usando IA. Fornece score calibrado de compatibilidade, identifica pontos fortes, gaps de habilidades e ações práticas para melhorar candidatura. Construído com Streamlit e Claude API.
Instalação
npx convez-ai stepmn818/job-search-assistant1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.
Métricas
Casos de uso
- Avaliar objetivamente antes de aplicar para uma posição
- Identificar gaps de habilidades para focar desenvolvimento
- Otimizar CV com sugestões específicas para cada vaga
Veredito

Ferramenta bem-pensada que resolve problema real de candidatos: decisão de aplicação baseada em dados, não intuição. Implementação sólida (Streamlit + Claude com structured prompting), roadmap claro incluindo rastreador e motor de reescrita de CV.
README
A tradução para o português deste README ainda está na fila.
🎯 AI Job Search Assistant
A personal AI agent that analyses how well your CV matches a job description — built with Streamlit + Claude API.
Portfolio project demonstrating: AI integration, structured prompting, and workflow automation thinking.
Problem Statement
Job hunting is time-consuming and opaque. Most candidates apply to roles without a clear read on how well their CV actually matches the job description — they either apply to everything hoping something sticks, or talk themselves out of roles they're actually qualified for.
Manually comparing a CV to a JD is tedious, subjective, and easy to get wrong. You miss keywords, underestimate transferable experience, or fail to spot the one gap a recruiter will flag immediately.
What This Tool Does
Upload your CV and paste a job description. The tool analyses both and returns:
- Fit Score — a calibrated 0–100 score reflecting overall alignment, derived as a weighted rollup of four sub-scores (so the number is never a black box)
- Sub-Scores — 0–100 each on Skills (35%), Experience (30%), Domain (25%), and Education (10%), with a one-sentence rationale explaining which dimensions drove the total
- Top Strengths — where your experience directly matches what the role needs
- Skill Gaps — what's missing or under-evidenced in your CV relative to the JD
- Quick Wins — specific, actionable steps to improve your fit right now, whether that's a CV tweak, a certification, or a reframe
- One-Line Recruiter Summary — how a recruiter reading your CV would likely position you for this role
The goal isn't to game the application process. It's to help candidates make smarter decisions about where to focus their energy — and show up better prepared when they do apply.
Built With
| Layer | Technology |
|---|---|
| UI | Streamlit |
| AI | Anthropic Claude API — claude-haiku-4-5-20251001 |
| File parsing | pypdf, python-docx |
| Language | Python 3.10+ |
| Dev tooling | Claude Code CLI |
How to Run
Prerequisites: Python 3.10+, an Anthropic API key (console.anthropic.com)
# 1. Install dependencies
pip install -r requirements.txt
# 2. Add your API key — create a .env file in the project root
echo "ANTHROPIC_API_KEY=your-key-here" > .env
# 3. Start the app
streamlit run CV_Fit_Scorer.py
Project Roadmap
| Phase | Feature | Status |
|---|---|---|
| 1 | CV vs JD fit scorer | ✅ Done |
| 2 | Application tracker (SQLite) | 🔜 Next |
| 3 | CV tailoring engine (auto-rewrite bullets to match JD) | 📋 Planned |
| 4 | Automated job discovery agent (daily digest via email) | 📋 Planned |
Built with Claude Code — the
CLAUDE.mdfile in this repo gives Claude full context on project architecture and conventions for easy extension.