memory

Name: memory
Availability: InStock
Author: harnessprotocol

Search entities and observations in a membrain knowledge graph.

9|2|Updated Mar 7, 2026

Search Tags:#knowledge graph #search #mcp #memory #observations #episodes #bfs traversal

Ready to run

npx skills add https://github.com/harnessprotocol/harness-kit --skill memory-harnessprotocol

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: memory
Source: https://github.com/harnessprotocol/harness-kit/tree/main/plugins/membrain/skills/memory
Command: npx skills add https://github.com/harnessprotocol/harness-kit --skill memory-harnessprotocol

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

It helps you retrieve and manage knowledge that’s scattered across sessions by letting you search, trace connections, and add new observations to your membrain knowledge graph.

Core Features & Use Cases

Graph search and entity discovery: Find relevant entities and their observations for a topic so you can ground future work in prior context.
Concept tracing: Run BFS traversals to understand how concepts connect and what information paths matter.
Knowledge growth workflows: Add observations and capture episodes so your memory stays current without manual restructuring.

Quick Start

Ask for a grounded result by saying: /memory search authentication

Frequently Asked Questions about memory

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

FAQPage Schema

How do I search and retrieve past context from a knowledge graph in Claude Code?▼

You can search a knowledge graph in Claude Code by using the memory search command to find relevant entities and their observations, grounding your current work in prior context. This allows fast recall of scattered session data.

What is BFS traversal and how does it help with concept tracing in an AI memory graph?▼

BFS traversal in a memory graph runs breadth-first searches to trace how concepts connect, revealing information paths and relationships between entities. This explains relationship discovery across your stored observations.

Do I need a local HTTP server to use MCP tools for memory search and trace operations?▼

You need membrain MCP tools for search, add, and episode operations. Optionally, you can run the local HTTP server on localhost:3131 for trace and status operations, but core search works without it.

How do I add new observations and episodes to my knowledge graph without manual restructuring?▼

You can add new observations and capture episodes through knowledge growth workflows, incrementally updating your memory graph. This keeps your stored context current without requiring manual restructuring of existing entities.

Can I use this memory Skill to manage knowledge scattered across multiple CLI sessions?▼

Yes, the Skill retrieves and manages knowledge scattered across sessions by searching entities and their observations in a membrain knowledge graph. It is designed for interactive CLI-based memory workflows.

What is the best way to trace concept connections in a knowledge graph?▼

Running BFS traversals is the best way to trace concept connections in a knowledge graph, allowing you to understand how concepts connect and identify what information paths matter for your current context.