A structured markdown framework for creating AI-ready knowledge graphs through categorized observations and linked relations.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install memory-notes
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install memory-notes using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Memory Notes?
Memory Notes is a technical specification for creating well-structured markdown files that an AI agent can parse into a searchable knowledge graph. It moves beyond simple text files by enforcing a schema that includes YAML frontmatter, rich prose context, semantic observations, and typed relations. This approach ensures that information is not just stored, but is programmatically discoverable and contextually linked.
As part of the Openclaw Skills ecosystem, this skill empowers developers to build a persistent memory layer for their AI agents. By utilizing specific syntax for facts and connections, the agent can traverse complex topics, recall specific design decisions, and understand the lineage of project requirements through a sophisticated graph-based architecture.
Memory Notes Use Cases
- Documenting technical architecture decisions and their associated trade-offs for team transparency.
- Building a personal knowledge management system that an AI can query to provide context during coding sessions.
- Tracking project milestones and linking them to specific tasks, people, and requirements.
- Creating a centralized repository for research notes where semantic links help identify non-obvious connections between topics.
How Memory Notes Works
- The process begins by defining a note with YAML frontmatter, establishing a unique title and searchable tags.
- Developers write substantive prose in the body to provide the 'why' behind the information, which aids in semantic search retrieval.
- Key facts are distilled into the Observations section using the
[category] contentsyntax, making them atomic units of knowledge. - Notes are linked together in the Relations section using wiki-links to form directed edges in the knowledge graph.
- AI agents use tools like search_notes or build_context to navigate these notes via stable memory URLs and wildcard patterns.
Memory Notes Setup
To start using Memory Notes within your environment, ensure your AI agent has access to the Basic Memory toolset. You can create your first note using the following command structure:
# Example of creating a note via a CLI-integrated tool
write_note --title "Project Alpha" --tags "internal,docs" --content "# Project Alpha\n\nDetails here..."
Notes are automatically assigned a permalink based on their title, which allows them to be addressed via memory:// URLs immediately upon creation.
Memory Notes Data Schema & Taxonomy
The Memory Notes skill organizes data into a consistent four-part structure within each markdown file:
| Component | Purpose | Format |
|---|---|---|
| Frontmatter | Metadata & Identity | YAML blocks with title, tags, and type. |
| Body | Contextual Prose | Free-form Markdown including headers and descriptions. |
| Observations | Atomic Facts | Unordered list using -[category] content #tag. |
| Relations | Graph Connections | Unordered list using type [[Target Title]]. |
All notes are addressable via the Memory URL scheme (e.g., memory://directory/note-title), providing a stable identifier that persists even if the physical file is moved.
Memory Notes Advanced Features
- Wildcard Querying: Use patterns like
memory://docs/*to gather context from entire directories at once. - Semantic Categorization: Invent custom categories for observations to enable precise searching, such as
search_notes("[decision]"). - Relational Traversal: Programmatically follow links between notes to build a comprehensive context window for the LLM.
- Incremental Graph Updates: Perform targeted edits to specific sections like Observations or Relations without rewriting the entire document.
SKILL.md
Loading
Related Openclaw Skills
Memory Lifecycle
A specialized skill for managing entity status transitions and knowledge archiving within an AI agent's memory system.
AMiner Academic Data Search
A comprehensive academic research toolkit leveraging the AMiner Open Platform to analyze scholars, papers, institutions, and patents.
SearXNG Web Search
A privacy-focused metasearch skill that aggregates results from multiple engines via the SearXNG HTTP API for AI agents.
Truncus Email
A high-performance transactional email skill for delivering programmatic alerts, reports, and notifications with deterministic reliability.
PharmaClaw Toxicology Agent
A predictive toxicology agent that performs deep drug safety profiling and risk classification using RDKit molecular descriptors.
Pharmaclaw Cheminformatics
A professional-grade cheminformatics agent for 3D conformer generation, pharmacophore mapping, and retrosynthetic fragmentation.
Featured*
