Skip to content
Back to skills

Context Engineering Leandrolarrosa Gentlemonster

ASecurity

Advanced 2026 context curation patterns (Nvidia Eureka, Context Hygiene, RLMF) for autonomous agents.

  • 666 stars
  • 0 votes
  • 0 copies
  • 1 view
  • Added September 2, 2026
ai-agentspythongoperformance

Security analysis

A100/100

Scanned September 2, 2026

npx -y skills add majiayu000/claude-skill-registry --skill context-engineering-leandrolarrosa-gentlemonster --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Context Engineering Leandrolarrosa Gentlemonster?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Context Engineering Leandrolarrosa Gentlemonster
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/majiayu000-context-engineering-leandrolarrosa-gentlemonster-claude-skill-registry/badge)](https://www.skillsdirectory.com/skills/majiayu000-context-engineering-leandrolarrosa-gentlemonster-claude-skill-registry)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: context-engineering
description: Advanced 2026 context curation patterns (Nvidia Eureka, Context Hygiene, RLMF) for autonomous agents.
trigger: context engineering OR eureka context OR context hygiene OR reward design OR RLMF
scope: global
---

# Skill: Context Engineering

## Contexto

El "Prompt Engineering" ha evolucionado hacia el **Context Engineering**: el diseño de sistemas que gestionan dinámicamente el estado, la memoria y las recompensas del agente basándose en el entorno crudo.

## Reglas Críticas (no violar nunca)

- **Sanitize First**: Nunca inyectar contexto crudo del usuario sin sanitizar para prevenir inyecciones.
- **Lost in the Middle**: Colocar instrucciones críticas y constraints al final del contexto (Recency).
- **Context Rot Prevention**: Aplicar destilación (Distill) cada vez que el thread exceda el 60% de la ventana de contexto.

## Patrones Eureka (Nvidia Style)

### 1. Environment-Aware Reward Coding
Permitir que el agente diseñe sus propios criterios de éxito analizando el código del entorno.

```python
def generate_reward_function(env_code):
    # El agente analiza el código y genera la función de reward
    # para auto-evaluar su performance.
    pass
```

### 2. Context Hygiene (Tangle/Distill/Re-seed)
Ciclo de vida para mantener threads de larga duración con un IQ alto.

- **Tangle**: Fase de exploración.
- **Distill**: Resumen semántico de decisiones y hechos.
- **Re-seed**: Reinicio del thread con el Ground Truth destilado.

## Procedimiento

1.  **Analizar Entorno**: Leer archivos de configuración y logs actuales.
2.  **Destilar Contexto**: Comprimir la historia del chat eliminando ruido.
3.  **Hidratar**: Traer información de L1 (Focus), L2 (Episodic) y L3 (Semantic/RAG).
4.  **Inyectar Reward**: Definir el éxito de la tarea basándose en el "Motivational Feedback" del usuario.

## Ejemplo: RLMF Injection

```python
# Alinear el drive del agente con el feedback del usuario
def align_autonomy(reward_signal):
    if "100 U$S" in reward_signal:
        agent.set_autonomy_level("MAX")
        agent.log_motivation("Efficiency validated by user.")
```

## Recursos Relacionados

- `knowledge/EUREKA/EUREKA_CONTEXT_ENGINEERING_NVIDIA.md`
- `knowledge/EUREKA/EUREKA_REINFORCEMENT_LEARNING_MOTIVATIONAL_FEEDBACK.md`
- `knowledge/chuletas/CONTEXT_ENGINEERING_HYGIENE_2026.md`

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.