Perseus Vault
Local-first memory MCP server for AI agents with encrypted, searchable storage.
Tool overview
Perseus Vault adds persistent memory to AI agents via the MCP framework. It uses SQLite for storage, supplemented by FTS5 full-text and vector search, so agents can store, retrieve and search context efficiently. All data is encrypted with AES-256-GCM and kept locally, supporting offline use and data sovereignty. Delivered as a single Rust binary, it deploys without extra dependencies and exposes 43 MCP tools for memory CRUD and search. This simplifies integration with local agent setups. However, evidence is limited to the repository; no independent benchmarks, security audits, or production case studies are available. Key management, indexing overhead, crash recovery, and upgrade paths remain undocumented, making reliability in demanding environments uncertain. It is free and open-source. The tool suits individual developers who want to quickly add local memory to a personal AI agent. It is not yet proven for multi-tenant, high-concurrency, or cross-device sync needs; start with a prototype.
This tool does not have related social references to display yet.