Persistent memory tools arrive
Memori Labs launched an OpenClaw plugin to add persistent, structured memory to multi‑agent gateways, promising to reduce agent 'amnesia' by automatically capturing relevant context announced. OpenViking is also gaining traction as a practical tool to consolidate session memory and tackle context drift in production reported.
The Memori OpenClaw plugin wires into OpenClaw’s lifecycle at before_prompt_build nationaltoday.com to inject recalled memories and at agent_end nationaltoday.com to extract and persist post‑turn facts, and the release notes list support for OpenClaw v2026.3.2+ with installable package @memorilabs/openclaw-memori. nationaltoday.com Memori Labs positions Memori Cloud as a fully hosted, SQL‑native memory layer for production agents marketwatch.com and promotes both managed and self‑hosted deployment options under CEO Adam B. Struck. nationaltoday.com The integration emphasizes data hygiene: the plugin implements “bulletproof sanitization” to strip OpenClaw system metadata, timestamps, and thinking blocks to prevent context pollution github.com and the announcement highlights production‑ready observability hooks for recall/ingest metrics and tracing. nationaltoday.com OpenViking’s upstream repo is published under volcengine/OpenViking and shows roughly 10.2k GitHub stars, indicating rapid community traction and active commits across examples, CLI, and core modules. github.com OpenViking implements a “context filesystem” that unifies memory, resources, and skills into hierarchical files and directories—promoting swap‑in replacement for context backends in LangChain workflows with a single configuration change openviking.ai, and its session management subsystem provides automatic context tracking, compression, and memory consolidation for long‑running agent interactions. deepwiki.com From an operational perspective, Memori’s SQL‑native, observability‑first plugin offers a fast path for multi‑agent gateway deployments with clear install and API key flows nationaltoday.com, while OpenViking’s file‑system, self‑evolving model targets large‑scale context consolidation and hierarchical retrieval that platform teams can swap into existing LangChain agent orchestration. github.com
