skills.rest
dingdyan
dingdyanCOMMUNITY
Explore all skills by this author →
GitHub

elite-longterm-memory

Record working state and curate long-term memory across agent sessions.

Updated Apr 5, 2026
Search Tags:#memory#vector search#lancedb#long-term context#git notes#mem0#wal protocol
One-click install
npx skills add https://github.com/dingdyan/openclaw-workspace-v2 --skill elite-longterm-memory-dingdyan
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: elite-longterm-memory
Source: https://github.com/dingdyan/openclaw-workspace-v2/tree/main/skills/elite-longterm-memory
Command: npx skills add https://github.com/dingdyan/openclaw-workspace-v2 --skill elite-longterm-memory-dingdyan

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill prevents AI agents from losing important context between messages and sessions, reducing repeated mistakes and forgotten decisions.

Core Features & Use Cases

  • Write-Ahead Logging (WAL): Writes critical state to SESSION-STATE.md before responding to avoid context loss on crashes/compactions.
  • Semantic Vector Recall (LanceDB): Recalls relevant past memories using vector search for better continuity on ongoing work.
  • Structured Decision Storage (Git-Notes) + Human Archive: Preserves permanent decisions in a Git-notes knowledge graph while keeping curated, readable summaries in MEMORY.md and daily logs.
  • Optional Cloud Backup (SuperMemory) and Auto-Extraction (Mem0): Syncs memory across devices and extracts facts to cut token usage.

Quick Start

Ask your AI agent to run: npx elite-longterm-memory init, then create and start using SESSION-STATE.md as the live context it writes to before each response.

Frequently Asked Questions about elite-longterm-memory

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

FAQPage Schema
How do I persist agent context across sessions to prevent lost decisions?▼

To persist agent context across sessions, use write-ahead logging to record critical working state to SESSION-STATE.md before responding, preventing memory loss during crashes or compactions.

How does vector search recall past memories for long-term AI context?▼

Vector search recalls past memories for long-term AI context by indexing historical interactions in LanceDB, enabling semantic retrieval of relevant past decisions and facts during ongoing conversations.

What's the best way to store permanent AI decisions in a Git repository?▼

The best way to store permanent AI decisions in a Git repository is using Git-Notes to preserve a structured knowledge graph of decisions while keeping curated summaries in MEMORY.md.

Can I sync long-term agent memory across multiple devices?▼

Yes, you can sync long-term agent memory across multiple devices by enabling optional SuperMemory cloud backup integration to sync memory and ensure continuity across environments.

How do I reduce token usage when extracting facts from agent conversations?▼

To reduce token usage when extracting facts from agent conversations, integrate Mem0 for auto-extraction, which automatically curates facts and cuts down unnecessary token consumption.

Do I need YAML-defined metadata to use write-ahead logging for agent state?▼

Yes, YAML-defined discovery metadata is required to configure the skill, alongside a WAL-managed SESSION-STATE.md file to properly log and persist agent state before responses.

Related Skills

Explore More Software Engineering Skills→
SpillwaveSolutionsSpillwaveSolutions
46

project-memory

Initialize and maintain project memory files under docs/project_notes.

Official
Basic
kaifashraffkaifashraff

elite-longterm-memory

Persist AI agent context across sessions using layered WAL-based memory stores.

Community
Advanced
darblexdarblex

elite-longterm-memory

Manage AI agent memory with RAM, vector search, and knowledge graphs.

Community
Advanced
tursodatabasetursodatabase
23.8k

transaction-correctness

Explain Turso WAL, commits, and checkpointing for transactional correctness.

Official
Advanced