dream

Consolidate agent memories into a three-tier graph with vector indexing.

3|1|Updated Jun 24, 2026

Search Tags:#database management #agent #memory consolidation #memory model #graph vector

Ready to run

npx skills add https://github.com/spqian/dreamweave --skill dream-spqian

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: dream
Source: https://github.com/spqian/dreamweave/tree/main/skills/dream
Command: npx skills add https://github.com/spqian/dreamweave --skill dream-spqian

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires python, sqlite, sqlite-vec, and includes scripts (resource) and references (resource) and assets (resource) components.

What problem does it solve?

This Skill consolidates the memory of your agent into a structured graph, improving recall and reducing noise.

Core Features & Use Cases

Memory Graph Construction: Weaves a graph+vector memory model with three tiers: instinct, recall, and archive.
Forgetting Curve: Implements a forgetting curve to decay noise and keep durable facts.
Use Case: Enhance your agent's memory with this Skill, allowing it to better recall information across different contexts.

Quick Start

Run the 'dream' command to consolidate the agent's memory.

Frequently Asked Questions about dream

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

FAQPage Schema

How do I build a memory graph for my agent to improve long-term recall?▼

You consolidate agent memory into a graph by running a command that structures memories into three tiers: instinct, recall, and archive. This uses SQLite and sqlite-vec for graph and vector indexing.

How does a forgetting curve work for agent memory consolidation?▼

The forgetting curve decays memory noise over time to keep durable facts. This Skill implements the curve within its memory model to actively manage duplication and reduce irrelevant data.

Do I need Python and SQLite to implement a graph vector memory model?▼

Yes, you need Python, SQLite, and the sqlite-vec extension to run this graph vector memory model. These dependencies handle the underlying database management and vector indexing required for consolidation.

What is the best way to manage agent memory duplication across different contexts?▼

The best way to manage agent memory duplication is to consolidate memories into a structured graph with a forgetting curve. This approach actively manages duplication and filters noise to enhance recall.

Can I use SQLite-vec for vector indexing in agent memory management?▼

Yes, you can use SQLite-vec for vector indexing in agent memory management. This Skill requires SQLite and SQLite-vec specifically to handle database management and build the graph vector model.

How do I structure agent memory tiers for instinct, recall, and archive?▼

You structure agent memory into instinct, recall, and archive tiers by weaving a graph+vector memory model. This Skill automatically consolidates incoming information into these three distinct levels for long-term recall.