skills.rest
Akanoa
AkanoaCOMMUNITY
Explore all skills by this author →
GitHub

Moerae - AI Memory

Store and retrieve agent-authored memories via semantic search across conversation and project scopes.

1|1|Updated Apr 8, 2026
Search Tags:#SQLite#embeddings#persistence#vector search#agent workflows#semantic retrieval#local memory
One-click install
npx skills add https://github.com/Akanoa/moerae --skill moerae-ai-memory-akanoa
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: Moerae - AI Memory
Source: https://github.com/Akanoa/moerae/tree/main
Command: npx skills add https://github.com/Akanoa/moerae --skill moerae-ai-memory-akanoa

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Moerae removes manual file-based indexing and context-window juggling by turning agent-written facts into a queryable local memory system that can be searched later.

Core Features & Use Cases

  • Auto-embedded storage with lifecycle: Store content with put and let Moerae manage segments, forgetting, promotion to project scope, and efficient retrieval.
  • Conversation + project scopes: Keep memory isolated per conversation by default, while promoting useful items to the project for cross-conversation search.
  • Large-content handling: If content is too long to embed directly, Moerae indexes only an agent-provided metadata summary while still allowing full retrieval later with get.
  • Use Case: An agent logs many tool interactions during a long session, then later answers “What did we decide about deployment keys?” by searching the conversation first and then the project-scope memory for promoted items.

Quick Start

Store a fact and retrieve it later by asking an assistant to run: moerae put -p myproject "The API rate limit is 1000 req/min" then moerae search -p myproject "rate limit".

Frequently Asked Questions about Moerae - AI Memory

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

FAQPage Schema
How do I store and retrieve agent memory locally using SQLite and vector search?▼

You can persist agent memory locally by writing facts with a put command and later retrieving them via semantic vector search across conversation and project scopes, eliminating manual file indexing.

How does semantic retrieval work for long content payloads in agent workflows?▼

For semantic retrieval of long payloads, the system indexes only an agent-provided metadata summary while still allowing full content retrieval later, bypassing direct embedding limitations.

Can I isolate agent memory per conversation and promote it to a project scope?▼

Yes, agent memory is isolated per conversation by default, but you can promote specific items to a project scope to enable cross-conversation semantic search and shared context.

Do I need to manually manage context windows or indexing for local agent memory?▼

No, you do not need manual indexing or context-window juggling. The system auto-embeds stored content, manages segments, and handles deterministic persistence and eviction behavior automatically.

What are the limitations of using local embeddings for agent memory persistence?▼

Local embeddings require configurable token thresholds, meaning excessively long content cannot be embedded directly and must rely on metadata-only indexing for successful semantic retrieval.

Related Skills

Explore More Software Engineering Skills→
aarushlohitaarushlohit

chromadb

Store, search, backup, and reset ChromaDB vector database entries via Python scripts.

Community
Advanced
shaprisshapris

answering-natural-language-questions-with-dbt

Generates and executes SQL to answer natural-language business questions using dbt semantic layer or compiled SQL.

Community
Advanced
Moerae-AIMoerae-AI
5

Moerae - AI Memory

Store and retrieve agent memories via embedded vectors and HNSW search.

Official
Intermediate
anthony-chaudharyanthony-chaudhary
30

ctx-overlay

Compact the active context window and prune the memory store.

Community
Basic