Reflexion
插件 警告 活跃Collection of commands that force LLM to reflect on previous response and output. Based on papers like Self-Refine and Reflexion. These techniques improve the output of large language models by introducing feedback and refinement loops.
To significantly enhance the quality, predictability, and accuracy of LLM outputs by integrating advanced reflection, critique, and memory update mechanisms.
功能
- Automatic reflection triggered by keywords
- Commands for self-reflection, critique, and memory curation
- Integration with Agentic Context Engineering (ACE)
- Multi-agent debate and LLM-as-a-Judge for critique
- Session data persistence for hook analysis
使用场景
- Improving code generation quality through iterative refinement
- Ensuring LLM outputs strictly adhere to requirements
- Creating evolving knowledge bases for AI agents
- Automating quality assurance for LLM-generated content
非目标
- Automatically fixing identified issues without user/LLM approval
- Providing generic LLM features not related to output improvement
- Replacing the core LLM's reasoning capabilities
License
- critical:License usabilityThe plugin is licensed under GPL-3.0, which is a strong copyleft license that may restrict its use in commercial or proprietary software without careful consideration and compliance.
安装
请先添加 Marketplace
/plugin marketplace add NeoLabHQ/context-engineering-kit/plugin install reflexion@context-engineering-kit包含 3 个扩展
Skill (3)
Comprehensive multi-perspective review using specialized judges with debate and consensus building
Curates insights from reflections and critiques into CLAUDE.md using Agentic Context Engineering
Reflect on previus response and output, based on Self-refinement framework for iterative improvement with complexity triage and verification
质量评分
警告类似扩展
Memory Bridge
100在 Cowork 项目之间共享记忆。停止教 Claude 同样的事情两次。
Si
100Self-Improving Agent: curate auto-memory, promote learnings to CLAUDE.md and rules, extract proven patterns into reusable skills. Provides /si:review, /si:promote, /si:extract, /si:status, and /si:remember slash commands.
Cc Safe Setup
99734 个用于 Claude Code 的安全钩子 — 在自主 AI 编码会话期间防止文件删除、凭证泄露、git 灾难和令牌浪费
Claude Code Hooks
99为 Claude Code 的自动化运行提供生产环境安全钩子。包括上下文监控、语法检查、分支保护、活动日志记录等。
Tdd
99TDD workflow enforcement with auto-test hook
Claude Recap
99为 Claude Code 提供基于主题的自动记忆——跨会话或压缩时绝不会丢失上下文。