ClawMem FAQs

Question 1

What is ClawMem and what problem does it solve for AI agents?

Accepted Answer

ClawMem is an on-device context engine and memory system for AI agents, specifically designed for Claude Code and OpenClaw. It solves the problem of transient agent memory by providing a persistent, evolving, and automatically surfaced knowledge base, all without cloud dependencies or API keys.

Question 2

How does ClawMem ensure AI agents always have relevant context?

Accepted Answer

ClawMem employs a sophisticated hybrid RAG system combining BM25, vector search, RRF, and cross-encoder reranking. It also uses multi-graph traversal (semantic, temporal, causal) and intent classification to surface the most relevant context, capture decisions, and generate handoffs for seamless session continuity.

Question 3

What are the primary integration options for ClawMem?

Accepted Answer

ClawMem offers two main integration paths: via Claude Code hooks and an MCP server, or as a native OpenClaw ContextEngine plugin. Both methods share a unified local SQLite vault, allowing decisions captured in one runtime to be immediately available in the other.

Question 4

Does ClawMem require cloud services, API keys, or internet access?

Accepted Answer

No, ClawMem is designed for fully local-first operation. It runs entirely on-device, utilizing local GGUF models and SQLite, eliminating any need for API keys, cloud services, or constant internet access. This ensures privacy, security, and full control over your agent's memory.

Question 5

How does ClawMem's memory system 'evolve' over time?

Accepted Answer

ClawMem features 'self-evolving memory notes' that automatically enrich documents with keywords, tags, and inferred causal links. It learns what matters through a feedback loop, boosts co-accessed documents, detects contradictions, and dynamically adjusts document durability based on revision counts and usage, creating an intelligent, adaptive knowledge base.

ClawMem

ClawMem

主な機能

ユースケース

主な機能

ユースケース