How does the Remember skill store information?

It primarily uses a local knowledge graph to store entities and relationships. This graph-first approach is zero-config and works out of the box without external dependencies.

Can I sync my memories across different machines?

Yes, by using the --mem0 flag and providing a MEM0_API_KEY, you can sync your graph data to the mem0 cloud for cross-device access.

Does it extract relationships automatically?

Yes, the skill parses your input for patterns like 'X uses Y', 'X recommends Y', or 'X requires Y' and automatically maps them as formal relations in the knowledge graph.

Can I scope memories to specific AI agents?

Absolutely. You can use the --agent flag to scope memories to a specific agent ID, ensuring that specialized knowledge is only invoked when that specific agent is active.

What is an anti-pattern in the Remember skill?

An anti-pattern is a decision or code structure marked with the --failed flag. This tells Claude that the approach caused issues and should be avoided or replaced with a better alternative.

Knowledge Graph & Pattern Memory

Name: Knowledge Graph & Pattern Memory
Author: yonatangross

byyonatangross

•

生产力与工作流

Persists architectural decisions, coding patterns, and lessons learned in a searchable knowledge graph for future AI sessions.

The Remember skill enables Claude to build a persistent project-specific knowledge base by storing success and failure outcomes as structured entities and relationships. By utilizing a graph-first architecture, it captures deep context about technology choices, design patterns, and agent-specific preferences without requiring external cloud dependencies. This persistent memory ensures that hard-won lessons and architectural constraints are respected throughout the development lifecycle, effectively creating a living best-practice library that evolves with your codebase.

主要功能

01Pattern Tracking: Explicitly marks patterns as success or anti-patterns to guide future AI decision-making.

02Automatic Entity Extraction: Detects technologies, agents, and relationship types like 'USES' or 'REQUIRES' from plain text.

03Flexible Scoping: Supports project-specific, agent-specific, or global cross-project memory storage.

04Graph-First Architecture: Stores entities and relations locally in a zero-config knowledge graph for instant retrieval.

0569 GitHub stars

06Cloud Sync Integration: Optional semantic cloud storage via mem0 for cross-device memory synchronization.

使用场景

01Standardizing coding conventions and pagination strategies across a multi-agent development workflow.

02Recording anti-patterns and known blockers discovered during testing to ensure the AI avoids repeating past mistakes.

03Documenting architectural decisions (e.g., choosing PostgreSQL over MySQL) to prevent redundant debates in future sessions.

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add yonatangross/orchestkit remember

For use in Claude.ai and ChatGPT

Download Skill