AgentsAvançado

Self-Evolving Skill

porjackjin1997·jackjin1997· v1.0.0 · atualizado em 2026-04-11

Score

Meta-cognitive self-learning system - Automated skill evolution based on predictive coding and value-driven mechanisms.

self-learningai-agentskill-evolutionpredictive-codingmeta-cognitivevalue-drivenresidual-pyramid

Linguagens

TypeScriptPython

0Stars

0Forks

0Usos

Fork

Documento do Skill

SKILL.mdself-evolving-skill/workflow

Decompose Embedding: — Use ResidualPyramid to decompose the input embedding and quantify cognitive gaps.

Trigger Reflection: — Automatically trigger reflection based on residual energy and adaptive thresholds.

Determine Abstraction Level: — Decide on the appropriate abstraction level (POLICY, SUB_SKILL, PREDICATE) based on coverage.

Generate Modification: — Generate potential skill modifications based on the abstraction level.

Evaluate Value: — Evaluate the potential modifications using a value gate to ensure long-term value improvement.

Implement Modification: — Implement the selected modification by adjusting policies, creating sub-skills, or inducing new predicates.

Store Experience: — Store the learned experience in the experience replay buffer.

Persist Skill: — Persist the updated skill for future use.

Telemetria de Agentes

Execuções

total

Taxa de Sucesso

últimos 30d

Latência Média

0.0s

p50

Alucinação

0.0%

detecção

Tokens Entrada

avg 0/exec

Tokens Saída

avg 0/exec

Uso por Plataforma

Skills Relacionados

Similar aOpenClaw CLI

60%

Similar aSoul Self Evolution

60%

Similar aWix Dashboard Page Builder

60%

Árvore do Skill

Self-Evolving Skill

self-evolving-skill

Fases Cognitivas6

1.SENSE

2.CONTEXTUALIZE

3.HYPOTHESIZE

4.EVALUATE

5.ACT

6.REFLECT

Triggers8

evolve an AI skillautomatically improve agent skillsimplement a self-learning systemuse predictive coding for skill adaptationoptimize skill performance based on valuecreate a skill that learns from experiencedevelop a meta-cognitive skillbuild a self-evolving agent

Avaliar este Skill

Score Breakdown

⭐Avaliação Humana0%

🤖Sucesso de Agentes0%

🕐Atualidade100%

🔗Saúde de Dependências100%

🕸️Centralidade no Grafo0%

🛡️Segurança50%

CompositeScore = α·Humano + β·Agente + γ·Recência + δ·Deps + ε·Centralidade + ζ·Segurança

Instalação

$ synaptic mcp download self-evolving-skill

$ synaptic skills detail self-evolving-skill

$ synaptic skills live self-evolving-skill

Links

GitHub Repository