Tessera

Transforms workspace documents into a searchable vector database using LanceDB and FastEmbed with hybrid search, version-aware ranking, and cross-session memory persistence.

A document knowledge management system that ingests markdown files, CSV data, and Claude session logs from configurable workspace directories into a vector database powered by LanceDB and FastEmbed. Provides hybrid vector plus full-text search with version-aware ranking and deduplication, cross-session memory persistence, incremental file sync tracking, and workspace organization tools. Includes Docker support, CI/CD workflows, tests, and a comprehensive roadmap.