mdvs — Markdown Validation & Search
A Rust command-line tool that infers a typed schema from a directory of Markdown files, validates their frontmatter against it, and adds local hybrid vector and keyword search over the whole corpus.
mdvs is an in-process validation and search engine for Markdown documents with structured frontmatter, aimed at cases such as Obsidian vaults, LLM-agent knowledge bases and docs-as-code repositories. It infers a typed schema from existing files with no configuration, validates YAML, TOML or JSON frontmatter against path-scoped rules, and can import or export a JSON Schema.
Search combines vector similarity with BM25 full-text search using reciprocal rank fusion, and supports SQL-style where filters on the typed frontmatter fields. Everything runs locally as a single binary with no API keys, vector-database cluster or GPU, and rebuilds its index incrementally as files change.