Engrava 0.5.0: a first-class MCP server
Engrava 0.5.0 pulls the MCP server out of the library. Install engrava now and you get the memory library and nothing else. The server ships separately, as engrava-mcp . The two were never really the same kind of thing — one is a dependency you import, the other is a process your MCP client spawns — and keeping them in one package meant everyone who installed the library also pulled in the…
Engrava 0.5.0 brings a standalone MCP server to the library. Previously, the memory library and server came bundled together, causing unnecessary dependencies for users who didn't require the server. Now, installing Engrava alone provides only the memory library, while the server stands independently as engrava-mcp. This change simplifies the installation process and ensures each component has its own clean install surface.
If you're building directly on the Python API, there's no need to change your setup, as installing Engrava with pip will remain the same: pip install engrava. However, if you want to use Engrava as a memory server for an MCP client, you'll need to install the standalone package engrava-mcp instead: uvx engrava-mcp or pip install engrava-mcp.
The server is a native stdio Model Context Protocol server that supports read tools, optional write tools, Engrava resources, and guided prompts for any MCP client capable of using stdio. It's listed in the official MCP Registry as ai.sovantica/engrava, so clients can easily discover it without manual setup.
Interestingly, while the library version remains engrava 0.5.0, the server version is engrava-mcp 0.5.1, meaning they follow different versioning schedules. The server's version is updated independently, typically every patch release.
Users of MCP will notice a few changes in version 0.5.0. The server now operates as a separate package, and installations and launch commands for MCP clients have shifted. For example, instead of pip install engrava[mcp], users now use pip install engrava-mcp or uvx engrava-mcp. Despite these changes, Engrava's library remains compatible, and the server still reads the same engrava.yaml or database path.
Additionally, Engrava 0.5.0 introduces several enhancements to the library itself, such as scoped ranked retrieval, audit verification, typed provenance capture, and reliability fixes. These improvements address issues like metadata and visibility filters in retrieval paths, hash-chain journal verification, and proper behavior of sqlite-vec backend and hybrid-fusion edge cases.
Engrava also offers opt-in lifecycle features like consolidation activation, deterministic memory hygiene, and caller-side whole-turn assembly, which are currently disabled by default.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
