Engrava is a lightweight, local memory layer for AI agents built on SQLite. It combines graph memory, full-text search, vector similarity, recency, and priority signals into a single hybrid recall interface with no server or per-operation metering required. The core is deterministic and avoids placing LLMs in the write path. It is installed via pip, requires only two primary calls (remember and recall), and includes an optional tamper-evident journal for verifiable history.
Engrava is an AI & ML product. It focuses on managing structured, searchable, and reliable long-term memory for AI agents without relying on heavy external databases or non-deterministic LLM writes. Engrava is an open-source project aimed at AI agent developers. The project is open source (MIT). It runs on the command line, embeddable surfaces, and API, and it can be self-hosted.
It is developed by Engrava, and the product first shipped in 2026. The project is developed in the open on GitHub with 197 commits in the last 90 days. Among its 6 catalogued features are Graph Memory, Hybrid Search, and Deterministic Consolidation. It exposes integrations via a public API.
Latest indexed changes and source events
Action Records: Memory for Things an Agent Did verified by the PulseGate indexer
Other apps tracked under the same category.
