Back to directory

Uteke

A local-first memory engine for AI agents, distributed as a single Rust binary, that stores and semantically recalls memories with no cloud service, no API keys and no language-model calls on the read path.

Uteke

Uteke stores short notes, such as a reminder to deploy to staging at a given time, and retrieves them later by meaning as well as by keyword, using on-device embeddings so recall involves no external API call and reportedly returns results in around 45 milliseconds once warm. It works with any MCP-compatible agent, including Claude Code, Cursor and Hermes, wired in through a single line naming its MCP command in the client's configuration file.

An onboarding command detects the user's setup, asks which agent they use, and configures the integration automatically. On the LongMemEval-S benchmark, which hides facts an agent needs across roughly 115 chat sessions per question, the project reports 98.4 percent recall in the top five results using zero language-model calls in the retrieval path, compared against a competing memory tool and a plain BM25 keyword baseline on the same question set. It installs through Homebrew, Cargo, Docker, or a downloaded binary for macOS, Linux or Windows.

More in Agent memory

See all 450
  • A long technical guide to AI agent memory: why stateless LLMs need it, the different memory architectures from vector stores to knowledge graphs, and how Cognee implements one as an open-source memory layer.

  • A local, file-based long-term memory layer for a personal AI that never summarizes conversations, keeping every line with a timestamp and searching by time first and words second, rather than retrieving only similar-seeming chunks.

  • NEOTH

    deepwiki.com

    A local-first personal AI daemon in Rust with five memory tiers, consent-gated tools, a WASM plugin sandbox, multi-provider model routing and signed audit logs, aimed at people who want both a daily assistant and a serious coding helper.