Show HN: Zerikai Memory: local code-memory MCP server, now with Jev (github.com)

The newly announced Zerikai Memory project introduces a local code-memory server designed to provide persistent, workspace-isolated memory for integrated development environments (IDEs). By leveraging Tree-Sitter code parsing and ChromaDB vector storage, Zerikai Memory effectively captures and indexes essential code entities like functions and classes, enabling developers to maintain context across different sessions and IDEs. Its local-first approach minimizes costs associated with token usage, vital for developers relying on AI models for coding assistance. The recent integration of Jev, TypeSafe AI's judgment engine, enhances Zerikai's capabilities by analyzing the relevance of retrieved code passages and supplying a plain-text evidence report to assist in decision-making. Jev operates as a narrow AI layer alongside existing LLMs, focusing on ensuring the quality and relevance of the information provided. This significant upgrade enhances the relevance scoring and efficiency of code retrieval, streamlining the coding process and reducing overhead for developers. Overall, Zerikai Memory aims to transform the way developers interact with AI, making it easier to manage context and improve productivity without incurring high computational costs.