Codevira addresses the common problem of AI coding agents starting from scratch in every session, leading to wasted tokens and inconsistent decisions. It functions as a Model Context Protocol (MCP) server, integrating directly into any project to offer a shared, persistent memory for AI agents. By providing capabilities like a context graph, semantic code search, project roadmaps, and decision logging, Codevira dramatically reduces token overhead by equipping agents with a deep understanding of the project's history, current state, and established rules, enabling faster, more informed, and consistent development.
Características Principales
01Semantic code search across the codebase using natural language
02Context graph for source file metadata, rules, and dependencies
03Atomic multi-file changeset tracking and session resumption
045 GitHub stars
05Phase-based project roadmap tracking for AI agents
06Structured decision logging from past AI agent sessions
