Reflection
Facilitates lightweight reflection and differential diagnosis by detecting AI providers or using local models to store and manage short-term memories.
概要
The Reflection server is a core component of the LXD MCP Suite, designed to enhance learning experience design through tailored reflection and differential diagnosis. It intelligently detects available AI providers like OpenAI, Anthropic, Gemini, or Ollama, while also offering a robust offline mode that utilizes a lightweight local model. This server securely stores short, bounded memories per key, providing essential tools over standard I/O for managing reflections, asking questions, noting observations, and summarizing contexts, all while staying small, safe, and fully functional without network access.
主な機能
- Exposes MCP tools over stdio for reflection, inquiry, and summarization
- Provides optional tailoring and validation for other MCP servers
- Functions fully offline leveraging local memory capabilities
- Automatic detection of AI providers (OpenAI, Anthropic, Gemini, Ollama) with local model fallback
- Secure local storage of bounded memories per key in JSONL format
- 0 GitHub stars
ユースケース
- Providing offline, local memory support for context-aware applications
- Tailoring and validating interactions for other MCP servers within a cohesive suite
- Enhancing learning experience design with AI-powered reflection and diagnostic capabilities