Self-Improving Agent
ПродвинутыйproductivityМинимальный контекст: 64K
Records learnings, mistakes, and corrections to enable continuous improvement of AI agent behavior. Maintains a structured memory of failures, user corrections, outdated knowledge, and discovered better approaches. Reviews past learnings before executing important tasks to avoid repeating mistakes.
Варианты использования
- Recording command failures and their resolutions for future reference
- Tracking user corrections to avoid repeating wrong approaches
- Maintaining a knowledge base of project-specific patterns and preferences
- Reviewing past learnings before executing similar tasks
- Discovering and documenting better methods for recurring workflows
Пример промпта
A command just failed unexpectedly. Record this learning: Context: [what you were trying to do] Command/Action: [what failed] Error: [error message or unexpected behavior] Root cause: [why it failed] Correct approach: [what works instead] Prevention: [how to avoid this in the future] Store this in the learnings file and review relevant learnings before executing similar tasks in the future.
Рекомендуемые модели
Совместимые инструменты
claude-codekiroopencodeany
Модальности
Вход: text, code
→Выход: text, file
Похожие Skills
Автор
OpenModels Community