convez.ai
SkillAutomaçãoIntermediário
0xdeval

Leads Screening Skill

0xdeval/leads-screening-skill

Triagem automática de candidatos com geração de relatórios e ranking

Ver no GitHub
InicialPluginScore · de 0 a 100

Skill reutilizável para Claude que automatiza a triagem de candidatos contra uma vaga de emprego. Mantém um perfil ideal de candidato, pontua candidatos, gera relatórios concisos e classifica o pool de candidatos analisados. Funciona em Claude Code, Claude Cowork e OpenAI Codex com gerenciamento manual de arquivos.

Instalação

$npx convez-ai 0xdeval/leads-screening-skill

1 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

Estrelas
0
Forks
0
Acompanhando
0
Saúde
42%
Último commit
há 5 meses
Versão
v1.1.1
Issues abertas
0
Contribuidores
1

Casos de uso

  • Automação do processo de recrutamento e seleção de candidatos
  • Geração de relatórios de avaliação de candidatos contra descrição da vaga
  • Ranking automático de leads qualificados para outreach direcionado

Veredito

Skill bem estruturada para equipes de recrutamento que usam Claude. Reduz trabalho manual na triagem inicial, mas requer setup de pastas e intervenção humana para validação final dos rankings. Útil em pipelines de RH que já operam com Claude Code ou Cowork.

README

A tradução para o português deste README ainda está na fila.

Candidate Screening Skill

A reusable Claude skill for screening recruiting leads against a vacancy. It maintains an ideal candidate portrait, scores candidates, writes concise reports, drafts outreach for suitable candidates, and ranks the screened pool.

The skill works with Claude Code and Claude Cowork. File management remains manual: place the vacancy and lead files into the expected folders, then ask Claude to screen them.

Install Directly from GitHub

Repository URL:

https://github.com/0xdeval/leads-screening-skill

Claude Code

Install the repository as a Claude plugin marketplace, then install the skill:

claude plugin marketplace add 0xdeval/leads-screening-skill
claude plugin install candidate-screening@leads-screening-skill

Restart Claude Code. The skill can trigger automatically for candidate-screening requests or be invoked as /candidate-screening.

For project-only installation:

claude plugin marketplace add 0xdeval/leads-screening-skill --scope project
claude plugin install candidate-screening@leads-screening-skill --scope project

Claude Cowork

  1. Open Claude Desktop and go to Customize → Plugins.
  2. Choose Add marketplace.
  3. Enter https://github.com/0xdeval/leads-screening-skill.
  4. Install and enable Candidate Screening for Cowork.

If direct marketplace installation is unavailable in the installed Claude Desktop version, download candidate-screening-plugin.zip from the latest GitHub release and use Upload plugin.

The exact Customize labels can vary by Claude Desktop version. Upload candidate-screening-plugin.zip, not the raw skill ZIP or the full repository source ZIP.

OpenAI Codex with $skill-installer

Ask Codex:

$skill-installer install https://github.com/0xdeval/leads-screening-skill/tree/main/candidate-screening

Or run the installer helper with repository and skill path:

python3 ~/.codex/skills/.system/skill-installer/scripts/install-skill-from-github.py \
  --repo 0xdeval/leads-screening-skill \
  --path candidate-screening

Restart Codex after installation.

OpenAI Codex Plugin Marketplace

Install the repository as a Codex marketplace, then install the plugin:

codex plugin marketplace add 0xdeval/leads-screening-skill
codex plugin add candidate-screening@leads-screening-skill

Start a new Codex thread after installation.

Manual Installation

The canonical skill is the complete candidate-screening/ folder.

# Claude Code
cp -R candidate-screening ~/.claude/skills/candidate-screening

# OpenAI Codex
cp -R candidate-screening ~/.codex/skills/candidate-screening

Set Up a Screening Folder

Create a working folder with this structure:

your-screening-folder/
├── job-description.md
├── potential-leads/
├── candidates/                 # created/populated by Claude
├── candidate-portrait.md       # generated by Claude
└── .candidate-screening/       # generated tracking state

Manual workflow:

  1. Put the vacancy text in job-description.md.
  2. Move candidate resumes or profiles into potential-leads/.
  3. Open the working folder in Claude Code or grant it to Claude Cowork.
  4. Ask: Screen all new candidates for this vacancy.
  5. Review reports under candidates/<candidate-name>/report.md.
  6. Add more files to potential-leads/ whenever needed and run screening again.

Previously reviewed, unchanged leads are skipped. Changed lead files become eligible for screening again.

Example Prompts

Screen all new candidates for this vacancy.
Who is the best candidate for the current role? Refresh anything stale first.
The job description changed. Refresh the candidate portrait and screen eligible leads.

PDF Support

Claude Cowork and recent Claude Code versions may read PDFs directly. If local extraction is required, install Poppler (pdftotext/pdftoppm) or Python packages such as pypdf or pdfplumber.

Build Release ZIPs

Run:

./scripts/package-skill.sh

The archives are written to:

  • dist/candidate-screening-plugin.zip: upload this to Claude Cowork. It contains .claude-plugin/plugin.json at the archive root.
  • dist/candidate-screening-skill.zip: raw skill folder for manual installation or inspection.

Repository Discovery Metadata

This repository includes:

  • .claude-plugin/marketplace.json and the wrapper's .claude-plugin/plugin.json for Claude Code and Claude Cowork.
  • .agents/plugins/marketplace.json for OpenAI Codex.
  • candidate-screening/SKILL.md as the canonical cross-platform skill.
  • plugins/candidate-screening as the Anthropic/OpenAI marketplace plugin wrapper.

The marketplace wrapper contains a synchronized copy because installed plugins do not preserve skill symlinks. Run ./scripts/check-plugin-sync.sh before publishing.

Privacy

Do not commit real resumes, candidate reports, job descriptions, or .candidate-screening/manifest.json. This repository's .gitignore excludes those local files by default.

Development

Run the tests:

python3 -m unittest discover -s candidate-screening/tests -v

License

MIT

Ver o original no GitHub

Bom para

Relacionados