memory-compass

Name: memory-compass
Availability: InStock
Author: wbin0001

Store and retrieve persistent AI memories with vector search and write-ahead logging.

4|1|Updated Mar 17, 2026

Search Tags:#memory #checkpoints #lancedb #wal #ai-memory #unified-search

Ready to run

npx skills add https://github.com/wbin0001/Memory-Compass --skill memory-compass

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: memory-compass
Source: https://github.com/wbin0001/Memory-Compass/tree/main
Command: npx skills add https://github.com/wbin0001/Memory-Compass --skill memory-compass

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) components.

What problem does it solve?

Memory Compass stores and retrieves persistent AI memories with fast semantic search, reliable write-ahead logging, and checkpoint-based recovery.

Core Features & Use Cases

High-speed semantic memory through LanceDB with cross-session recall
File-system WAL ensures durability and traceability of context
Unified search routes queries across memory tracks for robust results
Real-world use: AI agents need to remember task progress and decisions across sessions

Quick Start

Save a checkpoint for the current task and recover it later.

Frequently Asked Questions about memory-compass

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

FAQPage Schema

How do I add persistent memory to AI agents for cross-session context recall?▼

Persistent AI memory stores and retrieves context across sessions using LanceDB for fast semantic vector search. This enables agents to remember task progress and decisions without losing state between interactions.

How does checkpoint-based recovery work for long-running AI tasks?▼

Checkpoint-based recovery saves task state at specific intervals, allowing long-running AI tasks to resume from the last saved point after a failure. This prevents complete context loss and reduces redundant computations.

Why does my AI agent lose memory context during long-running operations?▼

Memory context is lost without durable storage mechanisms. A file-system write-ahead logging (WAL) ensures traceability and durability, recording context changes before applying them to prevent data loss during faults.

What is the best way to implement unified search across multiple AI memory tracks?▼

A unified search interface routes queries across multiple memory tracks, aggregating results for robust retrieval. This approach ensures comprehensive semantic searches across all stored AI memories.

Can I use LanceDB for high-speed semantic memory retrieval in AI applications?▼

Yes, LanceDB enables high-speed semantic memory retrieval by indexing AI memories as vectors. It allows fast cross-session recall by matching query embeddings against stored memory vectors.

Does write-ahead logging guarantee fault-tolerant memory operations for AI workflows?▼

Write-ahead logging (WAL) provides fault tolerance by recording memory operations to a file-system log before execution. If a failure occurs, the log replays uncommitted operations, ensuring reliable memory restoration.