Skip to content

Habilidades

Habilidades de IA jurídica: el directorio abierto de Agent Skills para el trabajo legal

Habilidades de IA legal seleccionadas de Anthropic, OpenAI, Lawvable, Skala y la comunidad. Todas gratuitas, y se instalan en Claude, ChatGPT o Gemini en menos de dos minutos.

How Agent Skills work

A skill is a folder, not a product. It holds a SKILL.md file - a name, a description of when to use it, and the instructions themselves - plus any scripts, templates or reference documents the task needs. Nothing inside is tied to one model, which is why the same folder works in more than one assistant.

  1. 01

    It stays out of the way

    The assistant only reads the name and the one-line description until something matches. A folder of fifty skills costs almost nothing to keep installed.

  2. 02

    It loads when it fits

    When your request matches a skill's description, the full instructions load - and only then. This is why a skill's description matters more than its length.

  3. 03

    It brings its own files

    A skill can ship a checklist, a clause bank, a script or a template, and use them while it works. That is the difference between a skill and a saved prompt.

Claude popularised the format and ships the runtime most people meet it through, so many skills here are written and tested against Claude. The format itself is plain files, and assistants that read it can run them - which is why this directory is organised by what a skill DOES, not by which assistant you use.

Explorar el directorio

Área de práctica

Mostrando 48 de 338 habilidades

Explorar el directorio completo en Lawvable

Browse by practice area

Every practice area with five or more skills has its own page, grouped by whether the skill knows the law or handles the file.

Cómo funcionan las habilidades de Legal AI

Una vez añadida una habilidad, cada solicitud se ejecuta de la misma forma dentro de HAQQ.

1

Elige una habilidad

Selecciona una del directorio abierto

2

HAQQ carga el contexto

Sus instrucciones y archivos se cargan en el modelo

3

Se ejecuta en el chat

Pregunta en lenguaje natural y la habilidad se ejecuta

4

Obtienes el resultado

Un resultado redactado, revisado o analizado