neolata-mem

What This Skill Does

neolata-mem is a graph-native memory engine designed specifically for AI agents, providing a sophisticated layer of persistence that surpasses standard context windows. By integrating Zettelkasten-style linking, hybrid vector-keyword search, and a unique biological decay algorithm, it allows agents to store, retrieve, and evolve information over time. The engine treats memories as nodes in a graph, facilitating connections between disparate facts rather than simple linear storage. It is completely zero-dependency and operates entirely offline by default, ensuring privacy and local execution without the need for cumbersome external databases like Neo4j or cloud-native infrastructure. The system is engineered for developers who need robust long-term memory that handles complex conflict resolution and information consolidation automatically.

Installation

Install the skill directly into your OpenClaw environment using the following command: clawhub install openclaw/skills/skills/jeremiaheth/neolata-mem

Alternatively, you can add it to your project via npm: npm install @jeremiaheth/neolata-mem

This library is lightweight, requiring no extra infrastructure like Docker or Python. It is designed to be plug-and-play for any Node.js environment version 18 or higher.

Use Cases

Use neolata-mem to maintain long-term user preferences across disparate chat sessions, where standard context windows are wiped. It is ideal for complex research tasks where an agent needs to synthesize findings into a unified knowledge graph. Additionally, it excels in multi-agent environments where different agents need to query a shared, evolving pool of information while resolving factual contradictions as new data surfaces. Use it whenever you need 'smart' memory that prioritizes high-value, frequently accessed information while naturally letting obsolete data fade through decay.

Example Prompts

1. "Store the fact that I prefer technical documentation to be concise and code-focused in the memory graph and link it to my current project context."
2. "Search through your long-term memory for any contradictory statements regarding the API authentication flow and suggest an updated version based on the most recent facts."
3. "Summarize the last five episodes of our project history from your memory to help me get back up to speed."

Tips & Limitations

To maximize performance, enable local embedding providers like Ollama to keep data off the cloud. Be aware that this skill is not a replacement for a full-scale knowledge graph database like Neo4j; it is a specialized engine for agentic memory. Avoid using this if you require global collaborative storage across thousands of users, as it is architected for high-performance individual or team-scoped agent state management.