skills.rest
EverClaw
EverClawOFFICIAL
Explore all skills by this author →
WebsiteGitHub

memory-upgrade

Diagnose and repair OpenClaw memory search with local embeddings and hybrid retrieval.

23|5|Updated Feb 23, 2026
Search Tags:#indexing#recall#embeddings#memory#diagnostics#hybrid search#openclaw
One-click install
npx skills add https://github.com/EverClaw/EverClaw --skill memory-upgrade-everclaw
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: memory-upgrade
Source: https://github.com/EverClaw/EverClaw/tree/main/memory-upgrade
Command: npx skills add https://github.com/EverClaw/EverClaw --skill memory-upgrade-everclaw

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) components.

What problem does it solve?

This Skill fixes broken OpenClaw memory search when the agent returns empty results, fails to recall information across sessions, or has no embedding provider configured.

Core Features & Use Cases

  • Local embeddings: Enables on-device embeddings without API keys or external services.
  • Hybrid retrieval: Combines BM25 keyword search with vector search for stronger recall.
  • Memory indexing: Indexes session transcripts and memory files so past conversations become searchable.
  • Ranking improvements: Adds MMR diversity and temporal decay so results are less repetitive and more recent memories rank higher.
  • Operational workflow: Includes diagnose, configure, verify, organize, and tag scripts for a complete maintenance flow.
  • Use case: Use it after a fresh install, when memory_search is empty, or when the agent needs better long-term memory quality.

Quick Start

Ask the assistant to diagnose OpenClaw memory search, apply the local memory upgrade, restart the service, and verify that indexing and recall are working.

Frequently Asked Questions about memory-upgrade

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
Why does my memory search return empty results after a fresh OpenClaw install?▼

Memory search returns empty results when no embedding provider is configured. You can fix this by applying a local memory upgrade to enable on-device embeddings and index session transcripts without requiring API keys.

How do I configure local embeddings for OpenClaw memory search without API keys?▼

To configure local embeddings for memory search, run the diagnostic and configuration scripts to enable on-device embedding generation, allowing the system to build search indexes without relying on external API services.

What is hybrid BM25 and vector retrieval for memory indexing?▼

Hybrid BM25 and vector retrieval is a combined search mechanism that blends keyword matching with semantic vector search to improve cross-session recall and deliver stronger memory search results.

How do I fix weak cross-session recall in my AI agent's memory?▼

Fix weak cross-session recall by indexing session transcripts and memory files, then applying MMR diversity and temporal decay so recent memories rank higher and results are less repetitive.

Does OpenClaw memory search support local embedding models for offline use?▼

Yes, OpenClaw memory search supports local embedding models for offline use. The configuration scripts enable on-device embeddings, hybrid retrieval, and file watching entirely without API keys or external services.

Related Skills

Explore More Software Engineering Skills→
augustsclaugustscl
2

noise-reduction

Classify and remove non-conversational noise from OpenClaw session JSONL transcripts.

Community
Advanced
profbernardojprofbernardoj

memory-upgrade

Diagnose and repair OpenClaw memory search with local embeddings and hybrid search.

Community
Advanced
Stream-claimsStream-claims

segment-search

Search OCR chunks in OpenSearch using hybrid BM25 and vector methods.

Official
Intermediate
J4JoshuaJ4Joshua

ollama

Serve local language and embedding models offline on Apple Silicon.

Community
Advanced