shodh-memory is a lightweight, offline Rust binary that gives AI agents persistent memory without any LLM calls. It uses Hebbian learning rules, decay curves, and knowledge graphs to strengthen important memories and forget unused ones in microseconds. Designed for edge, robotics, and local-first applications, it integrates with LangChain, LlamaIndex, Claude Code, and provides a REST API and MCP tools while keeping all data on-device.
In the AI & ML space, shodh-memory takes a focused approach. AI agents forgetting information or requiring expensive, non-deterministic LLM calls for every memory operation. shodh-memory is an open-source project aimed at AI developers. The project is open source (Apache-2.0). shodh-memory is available on the command line, and it can be self-hosted.
Varun Anku builds and maintains shodh-memory, and the product first shipped in 2025. The project is developed in the open on GitHub with 239 stars and 352 commits in the last 90 days. Among its 6 catalogued features are Hebbian Learning, Decay Curves, and Knowledge Graphs. It exposes integrations via a public API and an MCP server.
Latest indexed changes and source events
varun29ankuS/shodh-memory verified by the PulseGate indexer
Other apps tracked under the same category.
