A high-performance utility library providing privacy-first redaction, local JSONL storage, and offline text embeddings for AI memory management.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install openclaw-memory-core
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 openclaw-memory-core 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 Openclaw Memory Core?
Openclaw Memory Core serves as the foundational utility layer for managing agentic memory within the Openclaw Skills ecosystem. It is designed specifically for developers who prioritize data sovereignty and security, offering a robust set of tools to handle sensitive information and long-term data persistence without relying on external databases or cloud-based embedding models.
At its heart, the library enables seamless integration between different Openclaw Skills by providing a standardized way to redact secrets, store facts or decisions in append-only JSONL files, and perform semantic searches using a lightweight, offline embedding engine. This ensures that your AI agents remain fast, private, and capable of operating in restricted or offline environments.
Openclaw Memory Core Use Cases
- Redacting sensitive API keys and cloud credentials before storing agent interaction logs.
- Implementing a lightweight, database-free memory store using local file-based storage.
- Generating offline text embeddings for semantic search without incurring API costs or latency.
- Managing transient memory items with automated expiration support for temporary agent tasks.
How Openclaw Memory Core Works
- Incoming data or logs are piped through the Redaction module where sensitive patterns like API keys or PEM blocks are replaced with safe placeholders.
- The HashEmbedder processes the cleaned text using a deterministic FNV-1a hash-based algorithm to create a vector representation.
- Memory items are categorized into specific kinds (facts, decisions, or notes) and appended to local .jsonl files for persistent storage.
- When a retrieval request is made, the system calculates the cosine similarity between the query embedding and stored items to find the most relevant context.
Openclaw Memory Core Setup
To integrate this core library into your development environment for Openclaw Skills, install the package via npm:
npm install @elvatis_com/openclaw-memory-core
Once installed, you can import the specific modules required for your memory plugin, such as the Redactor for security or the JSONLStore for local data management.
Openclaw Memory Core Data Schema & Taxonomy
The library organizes memory into a structured format within .jsonl files to ensure compatibility across different Openclaw Skills.
| Attribute | Description |
|---|---|
kind |
The classification of memory (e.g., fact, decision, doc, note) |
content |
The redacted text content of the memory item |
embeddings |
A 256-dimension vector generated via HashEmbedder |
expiresAt |
An optional timestamp field for managing data lifecycle |
metadata |
A flexible object for storing additional contextual properties |
Openclaw Memory Core Advanced Features
- Comprehensive secret detection covering AWS, Azure, Stripe, GitHub, and generic JWT/Bearer tokens.
- Deterministic offline embedding generation that requires no external API calls or large model downloads.
- L2 normalization for accurate cosine similarity search in local vector environments.
- Support for custom expiration logic to maintain lean and performant local memory stores.
SKILL.md
Loading
Related Openclaw Skills
Openclaw INWX
A comprehensive automation plugin for managing the INWX domain registrar lifecycle, including DNS, registration, and DNSSEC within the Openclaw Skills ecosystem.
openclaw-help
A security-first plugin that adds a customizable /help command to your agent while keeping private workflows and shortcuts safe from public repositories.
Openclaw GPU Bridge
A high-performance bridge for offloading BERTScore and embedding calculations to remote GPU infrastructure.
OpenClaw Docker Plugin
A powerful plugin for managing Docker environments and Compose projects through AI agent commands.
OpenClaw Memory Docs
A local-first, audit-friendly memory store plugin for OpenClaw that provides explicit control over documentation capture and semantic search.
OpenClaw Model Failover Plugin
A gateway-level plugin that automatically switches to fallback LLM models when primary APIs hit rate limits or encounter quota errors.
Featured*
