Repolode
Submit
CortexReach

CortexReach/memory-lancedb-pro

Long-term memory plugin for OpenClaw agents that stores preferences and decisions in LanceDB and recalls them in later sessions.

View on GitHubWebsite
Stars
4,452
Forks
730
Open issues
50
Last commit
yesterday

What is it?

memory-lancedb-pro is a memory plugin for OpenClaw agents. It captures facts, preferences, decisions and project context from conversations, stores them in a LanceDB vector database, and injects relevant memories before later replies so users do not have to repeat themselves across sessions.

Retrieval combines vector similarity with BM25 full-text search, followed by optional cross-encoder reranking through providers such as Jina, SiliconFlow, Voyage AI and Pinecone. Memories are organized into scopes such as global, per-agent, per-user and per-project, and a Weibull decay model lets rarely used memories fade while frequently recalled ones persist. Smart extraction uses an LLM to classify memories into six categories.

The plugin also ships a management CLI for listing, searching, exporting, importing, re-embedding and upgrading the memory store. It supports OpenAI-compatible embedding APIs and includes notes for CPU instruction requirements of native LanceDB search.

Key features

  • 01Hybrid retrieval combining vector search and BM25 full-text search.
  • 02Optional cross-encoder reranking with Jina, SiliconFlow, Voyage AI or Pinecone.
  • 03Automatic capture of memories from conversations and automatic recall before replies.
  • 04Multi-scope isolation for global, per-agent, per-user and per-project memory.
  • 05Weibull-based decay and tiered promotion for memory lifecycle management.
  • 06Management CLI for search, stats, export, import, re-embedding and upgrades.

Use cases

  • Giving a personal OpenClaw assistant memory of coding style and preferences across sessions.
  • Keeping a record of past technical decisions so an agent can explain why a choice was made.
  • Separating memories between multiple agents or projects on one gateway.
  • Running a local or self-hosted embedding model with an OpenAI-compatible endpoint.

Quick start

  1. 1

    Install the plugin via OpenClaw CLI

    openclaw plugins install memory-lancedb-pro@beta
  2. 2

    Add the plugin config to openclaw.json and set the memory slot

  3. 3

    Validate config and restart the gateway

    openclaw config validate && openclaw gateway restart

Things to know

  • Requires a CPU with AVX support; some Linux x64 builds may need AVX2 for native vector search.
  • Needs an embedding provider and API key, such as OpenAI or Jina, for semantic memory.
  • Depends on the OpenClaw host; it is a plugin, not a standalone tool.
  • Upgrading from older versions or changing embedding models requires careful re-embedding into a new database.

Topics

#ai-agent#memory#vector-database#lancedb#rag#hybrid-search#reranking#llm#openclaw#openclaw-agent

FAQ

What is memory-lancedb-pro?
Long-term memory plugin for OpenClaw agents that stores preferences and decisions in LanceDB and recalls them in later sessions. memory-lancedb-pro is a memory plugin for OpenClaw agents. It captures facts, preferences, decisions and project context from conversations, stores them in a LanceDB vector database, and injects relevant memories before later replies so users do not have to repeat themselves across sessions.
Is memory-lancedb-pro free to use? What license does it have?
memory-lancedb-pro has no declared license. Its code is public, but without a license you have no explicit permission to reuse it; ask the author before relying on it.
Is memory-lancedb-pro still maintained?
Actively maintained. The last commit to memory-lancedb-pro was on Oct 21, 2026.
What can I use memory-lancedb-pro for?
Giving a personal OpenClaw assistant memory of coding style and preferences across sessions. Keeping a record of past technical decisions so an agent can explain why a choice was made. Separating memories between multiple agents or projects on one gateway. Running a local or self-hosted embedding model with an OpenAI-compatible endpoint.
How do I install memory-lancedb-pro?
Install the plugin via OpenClaw CLI: openclaw plugins install memory-lancedb-pro@beta Add the plugin config to openclaw.json and set the memory slot Validate config and restart the gateway: openclaw config validate && openclaw gateway restart
Where can I get memory-lancedb-pro?
The source code and installation instructions are on GitHub: https://github.com/CortexReach/memory-lancedb-pro

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects