ClawGraph is a graph-based memory CLI that enables AI agents to store and recall durable user facts as entities and relationships.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install clawgraph
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 clawgraph 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 ClawGraph?
ClawGraph is a sophisticated graph-based memory tool that empowers AI agents to maintain state across multiple sessions. As a key component in the Openclaw Skills ecosystem, it allows developers to store explicit facts about users, projects, and relationships in a persistent knowledge graph. Unlike systems that rely on weak inferences, ClawGraph prioritizes high-fidelity data storage, ensuring that the information recalled is accurate and directly derived from user input.
By integrating this skill, agents can proactively build a detailed map of information that survives restarts and provides a rich context for complex decision-making. It functions as a local memory vault, storing data in the user's home directory while offering structured JSON output for seamless machine readability.
ClawGraph Use Cases
- Capturing long-term user preferences, roles, and project affiliations.
- Tracking evolving relationships between different entities in a business or technical domain.
- Providing a reliable source of truth for agents tasked with long-term research or management.
- Recalling historical context when a user asks about previously discussed topics.
How ClawGraph Works
- The AI identifies explicit, durable facts within a conversation that should be preserved.
- The fact is passed to the ClawGraph CLI using the add or add-batch command.
- The system decomposes the natural language input into entities and relationships using a language model.
- Data is persisted to a local knowledge graph using idempotent logic to ensure consistency.
- When needed, the agent performs natural language queries against the graph to retrieve relevant facts.
ClawGraph Setup
To get started with ClawGraph as part of your Openclaw Skills setup, ensure you have an OpenAI API key and use the following installation steps:
uv tool install clawgraph==0.1.3
export OPENAI_API_KEY='your_api_key_here'
You can further customize the model and database paths by editing the config file at ~/.clawgraph/config.yaml.
ClawGraph Data Schema & Taxonomy
ClawGraph organizes data into a structured graph format consisting of nodes and edges. The schema is observable through the ontology command.
| Component | Description |
|---|---|
| Entities | Nodes representing distinct objects, people, or concepts. |
| Relationships | Connections defining how entities interact (e.g., "works at", "leads"). |
| Knowledge Graph | The aggregate of all entities and relationships stored at ~/.clawgraph/data. |
| JSON Output | All data retrieval is formatted as structured JSON for easy integration. |
ClawGraph Advanced Features
- Batch fact ingestion for efficient processing of multiple data points in a single LLM call.
- Idempotent storage mechanism using MERGE logic to prevent data duplication.
- Full graph export functionality for auditing or migrating stored knowledge.
- Comprehensive Python API for developers building custom automation or complex agent workflows.
- Support for custom OpenAI-compatible endpoints to facilitate diverse model usage.
SKILL.md
Loading
Related Openclaw Skills
Context Usage Monitor
A real-time token tracking and context window management tool for monitoring AI session limits.
Efficiency Hub
A comprehensive daily task management center that unifies search, scheduling, and archiving into a single command interface.
Essential Skills Package
A versatile suite of foundational capabilities designed for high-efficiency academic research, software development, and system automation.
Literature Search
A professional academic literature search and analysis tool supporting multiple platforms including arXiv, Google Scholar, and PubMed.
Cross-Border E-Commerce Selection Tool
An automated suite for cross-border sellers to perform keyword research, calculate platform-specific profits, and generate AI-optimized listings.
SEO Audit Pro
Generate detailed technical SEO audits and keyword-focused content briefs automatically for any URL.
METADATA
Featured*
