MCPFast / Tools / Persistent code-knowledge memory for AI chats via local MCP
A local MCP server caching codebase structure and summaries in SQLite for fast, token-cheap queries, avoiding re-reading files.
View on GitHub→This tool provides a local MCP server that acts as a persistent memory for AI chat interactions, specifically focusing on codebase structure and summaries. By leveraging SQLite for caching, it enables rapid, token-efficient queries, eliminating the need for AI models to repeatedly re-read and re-process source files. This significantly speeds up development workflows and reduces the computational cost associated with AI-assisted code analysis and generation.
The core function of this tool is to build and maintain a local knowledge base of your codebase. It parses your project's files, extracts structural information, and generates concise summaries. This data is stored in an SQLite database. When you interact with an AI chat agent, the MCP server intercepts requests and queries this local database first. If relevant information is found, it's provided to the AI, bypassing the need to access and process the actual files. This results in faster responses and lower token consumption for the AI model.
This tool is designed for AI developers and engineers who are integrating AI models into their development workflows. It is particularly beneficial for those working with large codebases where repeated file parsing by AI can become a bottleneck. Developers looking to optimize AI inference costs, improve the speed of AI-assisted code reviews, or enhance the context awareness of AI agents in their projects will find this tool invaluable. It's a practical solution for building more efficient and cost-effective AI-powered development environments.